Normalisation Between Belgrade and Pristina: Binding Force and Legal Effects of the Brussels and Ohrid Agreements

Abstract

This article revisits the debate on whether the Brussels Agreement and the Ohrid Agreement, including its Implementation Annex, are legally binding. It develops a three-test framework that separates international-law binding force from EU-law legal effects. Tests A and B adapt the International Court of Justice (ICJ) indicators of animus contrahendi and acceptance through subsequent conduct, acquiescence and silence. Test C examines whether the agreements produce legal effects through EU enlargement conditionality, monitoring and reporting. The analysis finds that the treaty status of both instruments remains contestable. The Brussels Agreement is textually specific, yet intent signals are mixed, practice remains reversible and treaty-type obligation structures are weak. The Ohrid Agreement is drafted in a more treaty-like register, but references to a future “legally binding agreement” and the politics of non-signature leave inter se binding force unsettled. Nonetheless, both agreements can produce EU legal effects. They operate as enlargement benchmarks that shape assessments of Serbia’s and Kosovo’s progress, while Commission reporting and standardised compliance indicators may indirectly bind EU institutions through consistency, equal treatment and legitimate expectations.

1. Introduction

The dialogue on normalisation between Belgrade and Pristina under EU auspices has been ongoing since 2011. Among more than 30 Dialogue agreements, there are two central pieces: the Brussels Agreement (the First Agreement of Principles Governing the Normalisation of Relations) and the Ohrid Agreement (the Agreement on the Path to Normalisation and its Implementation Annex). While the enduring politics of non-implementation and incremental yet reversible progress have been analysed extensively (Bieber 2015; Economides and Ker-Lindsay 2015; Beysoylu 2018; Troncotă 2018; Bargués et al. 2024; Doli and Xhaferri 2025), the question of whether the Agreements are legally binding remains comparatively under-examined (Istrefi 2023; Đukanović and Vučić 2024). Existing discussions often slide between “bindingness” as political signalling and binding force as a legal category, and they rarely separate international-law indicia from EU legal effects within a single operational framework.

The Brussels Agreement is frequently framed—by domestic courts and political actors—as a political agreement rather than a treaty in the classical sense: an agreement “concluded between States in written form and governed by international law” (United Nations 1969, Vienna Convention on the Law of Treaties (VCLT), Art. 2(1)(a)). However, Kosovo1, an entity not universally recognised as a sovereign state, ratified the Agreement in its Assembly (Assembly of the Republic of Kosovo 2013). On Serbia’s side, while there was no classic treaty ratification, the National Assembly adopted the Government’s report on the negotiations with Pristina by a large majority (Narodna skupština Republike Srbije 2013), signalling institutional endorsement of the deal’s basic orientation. The Constitutional Court of Serbia treated the Brussels Agreement as a political deal (Constitutional Court of the Republic of Serbia 2014), while the Constitutional Court of Kosovo held that the “General Principles/Main Elements” for the Association/Community of Serb-majority municipalities (ASM/CSM) were “not entirely in compliance” with Kosovo’s legal framework (Constitutional Court of the Republic of Kosovo 2015). Complicating matters further, political representatives often prioritise domestic narratives over consistency, producing contradictory public statements (Kovačević and Surlić 2025). The EU has nevertheless treated the Brussels Agreement and previous Dialogue agreements as binding on the parties and relevant to Serbia’s and Kosovo’s EU path, calling on both sides to fully implement all previous Dialogue agreements (Council of the European Union 2023).

The controversy is even sharper in relation to the Ohrid Agreement. Albin Kurti, Kosovo’s Prime Minister, has sent mixed messages: he has called on the EU to “make the status of the annex legally binding” (Kosovo Online 2023), but has also stated that the Ohrid Agreement is already “legally binding” (KoSSev 2023). On the other hand, Serbia’s President Aleksandar Vučić declared that the Agreement was not signed and that he would not sign any “international legally binding documents” with Kosovo (Zimonjić Jelisavac 2023). For its part, the EU “recalled repeatedly that the [Ohrid] Agreement is binding in its entirety under international law”, yet Article 6 states that the Agreement “should lead to a legally binding agreement” (European External Action Service [EEAS] 2023a, Art. 6).

The confusion over whether the two arrangements are legally binding partially arises because “legally binding” carries two distinct meanings: binding force and legal effects. The Brussels and Ohrid Agreements cannot therefore be assessed through form alone. Their ambiguity lies in the interaction between political commitment, possible inter se obligation, domestic constitutional reception, and EU conditionality. This article asks under what conditions, if any, the two Agreements possess binding force under international law, and how they may produce EU legal effects even where treaty-type binding force remains contestable.

To address this question, the article brings the literature on Belgrade–Pristina normalisation into conversation with two additional bodies of work. The first concerns the international-law distinction between treaties, political commitments and non-treaty instruments (Chinkin 1989; Klabbers 1996; Aust 2013; Dörr and Schmalenbach 2018). This body of work is relevant because the absence of ordinary treaty form does not, by itself, resolve the question of binding force; what matters is whether the parties manifested an intention to create legal obligations, and whether subsequent conduct supports an inference of acceptance, reliance or acquiescence. The second concerns EU soft law and legal effects (Senden 2004; Stefan 2012; Terpan 2015; Eliantonio et al. 2023). It shows that instruments lacking binding force may nevertheless become legally relevant within the EU legal order when incorporated into interpretative practice, administrative assessment, enlargement benchmarks or conditionality mechanisms.

The article contributes to the literature on Belgrade–Pristina Dialogue by separating two legal questions often conflated in political and legal debate: whether the Agreements possess binding force under international law, and whether they produce legal effects within the EU legal order. It addresses these questions through a three-test framework. Tests A and B assess international-law binding force through indicia of intent, acceptance, reliance and subsequent conduct. Test C examines EU legal effects generated through enlargement benchmarks, monitoring and institutional self-commitment. The article first develops this framework, then presents the content of the two Agreements, and finally applies the tests to both agreements.

2. Analytical Framework: Binding Force and Legal Effects

This framework sets out two types of tests. First, Tests A and B draw on ICJ case-law indicators and assess binding force under international law. Second, Test C rests on EU soft law and examines how the EU can generate legal effects independently of international-law binding force. The distinction is methodological as well as substantive. ICJ case law is used only to identify evidentiary indicia of intent, acceptance and reliance when assessing whether the Agreements possess binding force inter se. Test C asks a separate question: whether, regardless of that international-law characterisation, the same Agreements become legally relevant within the EU’s enlargement, monitoring, reporting and conditionality structures.

2.1. Binding Force

Given the absence of ordinary treaty formalities, the Brussels and Ohrid Agreements are not hard law in the treaty sense. However, Article 3 VCLT makes clear that falling outside the Convention’s scope does not, by itself, affect the legal force of agreements involving other subjects of international law or agreements not in written form. Thus, non-treaty form is relevant, but not dispositive (Dörr and Schmalenbach 2018; Fitzmaurice and Merkouris 2020), and it does not preclude the possible emergence of binding force through non-treaty pathways, including soft-law mechanisms (Chinkin 1989; Abbott and Snidal 2000). Simultaneously, such agreements do not directly trigger pacta sunt servanda, state responsibility, or remedies for breach (Zimmermann and Jauer 2021). In the practice of the ICJ, illustrated here through Qatar v. Bahrain (1994), Nicaragua v. Honduras (2007) and Peru v. Chile (2014), what matters is animus contrahendi—the intention to create legal obligations—evidenced by words, contemporaneous statements, subsequent conduct, and surrounding circumstances. That is precisely why the line between a political understanding and a legal obligation is often drawn by how the parties behave before and after the arrangement. It is important to note that the ICJ cases below are used for a limited methodological purpose: they are not cited as substantive analogies to the Brussels and Ohrid Agreements (whose political setting is distinct), but to identify the ICJ’s indicators for binding force outside ordinary treaty form.

Test A. A locus classicus illustration of the Court’s intent-centred approach is Qatar v. Bahrain. The ICJ treated an exchange of letters (1987) plus agreed minutes (Doha 1990) as a binding title of consent to jurisdiction irrespective of form; it read the Arabic dual al-ṭarafān (“the two parties may submit…”) as permitting unilateral seisin, that is, unilateral referral of the dispute to the Court, prioritising animus contrahendi gleaned from text and context over formal labels (Mendelson 2001). This shows both the potential of soft instruments to bind and the drafting risk of ambiguous language; notably, the parties then litigated the merits and accepted the 2001 judgment, suggesting the operational merit of such consent (Wiegand 2012). Within that limited methodological use, Nicaragua v. Honduras and Peru v. Chile illustrate the importance of sustained implementation rather than episodic or administratively explainable conduct: while, in the former, the Court refused the agreement argument since oil/fisheries practice, patrols, and fishermen’s statements were episodic, administratively explainable, and did not evidence mutual intent to be legally bound (Bekker and Stanic 2007; Lathrop 2008b; Tanaka 2008), in the latter case, the Court accepted that an agreement existed based on the parties’ instruments (notably the 1954 Special Maritime Frontier Zone Agreement) and their operational implementations (Burney 2014; Infante Caffi 2014).

Since the ICJ treats form as non-decisive and instead assesses whether the parties intended to create legal relations, Test A translates animus contrahendi into observable dimensions. Manifestation: Is there a record (minutes, parallel notes, verbatim readouts) of mutual assent to specific obligations (who/what/when)? Animus contrahendi indicators: Is there language of obligation that refers to an agreement, coupled with modalities of acceptance? Reliance and performance: Are there co-ordinated, reciprocal steps taken shortly after that only make sense if both sides understood themselves as bound (e.g., permits issued, patrol patterns adjusted, payments paused/authorised), without timely, reasoned protest?

Test B. While the ICJ looks to established practices, such practice need not take the form of symmetrical reciprocity: subsequent conduct can serve as evidence of a shared understanding. In treaty contexts, such practice may be relevant to interpretation (United Nations 1969, VCLT art. 31(3)(b)); in the present analysis, it is used more modestly, because the inquiry does not presuppose an existing treaty to be interpreted, but treats subsequent conduct only as evidence of possible acceptance and reliance. The Court’s acquiescence jurisprudence is most developed in territorial title and boundary settings; it is used here only as an evidentiary template for when consent may, or may not, be inferred from the parties’ conduct or silence, not as a claim that territorial acquiescence necessarily generates treaty-type obligations. A particularly contested form of subsequent conduct is silence. In international law, silence is conceptually contested and context-dependent—ranging from permissive logic to completeness arguments, non liquet, and sui generis approaches (Quane 2014; Azaria 2025). In practice, however, courts are reticent to embrace non liquet and cautious of sui generis reasoning because of legal certainty concerns; they instead evaluate silence through good faith and the stabilising function of settled expectations. Accordingly, silence is treated as legally meaningful only where it responds to a precipitating act or claim that called for a reaction, the silent party knew (actually or constructively) and could object, and it remained silent for a reasonable time in a manner consistent with acceptance (Kopela 2010).

Two cases illustrate the conditions under which silence may support an inference of acquiescence. In Malaysia v. Singapore (2008), the ICJ accepted Johor’s (Malaysia) original title but held that, by 1980, sovereignty over Pedra Branca/Pulau Batu Puteh had passed to Singapore, locating the shift in derivative title grounded not in the weighing of competing effects but in tacit processes. The Court set Singapore’s post-1953 acts à titre de souverain against Malaysia’s prolonged non-protest and affirmative cartographic and reporting practice. These acts included investigations and permission requirements for Malaysian officials, state displays, installations and proposals. The Court concluded that the facts showed a “convergent evolution” culminating in Singapore’s title, thus confirming that, in specific circumstances, inaction may be interpreted as consent (Lathrop 2008a; Beckman and Schofield 2009). For the purposes of this article, the relevant equivalent is not à titre de souverain but authority-allocating conduct, understood as conduct concerning institutional authority rather than territorial sovereignty. In another case concerning the Temple of Preah Vihear (Cambodia v. Thailand 1962), the ICJ held that since Thailand failed to protest the Annex I map for over fifty years, despite knowledge and opportunity, and thus acquiesced to the frontiers from the map. The Court, therefore, found the Temple to be on Cambodian territory and ordered Thailand’s withdrawal (Uchkunova and Temnikov 2013). In its 2013 interpretation judgment, the Court clarified that “vicinity” refers to the Preah Vihear promontory (not the wider Phnom Trap area) and confined itself to interpreting and not revising the 1962 holding (Uchkunova and Temnikov 2013).

Test B operationalises the Court’s acquiescence logic as follows. Manifestation: Is there a clear claim or a pattern of acts allocating authority that called for a reaction, and was it known (actually or constructively) to the other side? Are there repeated, concordant acts attributable to the parties, rather than isolated gestures? Acceptance vs. tolerance indicators: Do the facts point to acceptance rather than mere tolerance or strategic non-response? Are there any objections or reservations? Reliance and performance: Has behaviour on both sides evolved into stable routines that only make sense if the claim was accepted (e.g., joint bodies, routine procedures, or institutional allocations operating without reservation)?

2.2. Legal Effects

At this juncture, it is important to address two related questions that fall outside the scope of this article. The first concerns the broader relationship between EU law and international law. This is a highly debated question: while the orthodox view treats EU law as subject to international law, developments over the past two decades suggest a more complex picture, in which EU law is both rooted in international law and increasingly understood as an autonomous legal order with its own constitutional logic (Ziegler 2016; Molnár and Wessel 2024; De Witte 2017). The second concerns whether the EU can effectively enforce bilateral normalisation commitments after accession (Basheska 2025). These questions are especially relevant to agreements such as the Brussels and Ohrid agreements, which sit at the intersection of international-law contestation and EU enlargement conditionality. For the purposes of this article, however, the more immediate issue is narrower: whether, before accession, the EU can translate such Agreements into legal effects through benchmarks, monitoring, reporting and funding conditionality. This is the function of Test C.

Test C. Even in cases where an arrangement lacks binding force in the international law sense, it may still produce legal effects within the EU legal order once it is incorporated into EU assessment structures that carry legal or administrative consequences, including enlargement benchmarks, monitoring arrangements, reporting practices and funding conditionality. Senden defines EU soft law as “rules of conduct that are laid down in instruments which have not been attributed legally binding force as such, but nevertheless may have certain (indirect) legal effects, and that are aimed at and may produce practical effects” (Senden 2004, p. 3). In this context, legal effects is an umbrella concept for instruments that lack binding force but can nonetheless shape legal reasoning and administrative practice (Senden 2004). Accordingly, legal effects are not the same as binding force. However, the Court of Justice of the European Union (CJEU) has held (Grimaldi) that soft-law instruments “cannot be regarded as having no legal effect” (Case C-322/88, Grimaldi, Summary; Stefan 2012, p. 2).

In the Commission’s practice (e.g., annual reports, guidance, methodological papers, and conditionality decisions), EU soft instruments can produce legal effects along a spectrum. At the lightest end, they can function as ordinary interpretative context for binding instruments: they are typically used to support and clarify the interpretation of hard-law rules rather than to serve as an autonomous legal basis (Senden 2004, p. 147; Jiménez 2022, p. 10). Because this effect is often too diffuse to do analytical work in a highly contested setting such as the Brussels and Ohrid Agreements, the analysis concentrates on the two stronger mechanisms. First, where a binding instrument makes implementation and/or progress legally relevant, legal effects serve as a mandatory interpretative aid by analogy with Grimaldi. In Grimaldi, the CJEU held that recommendations are not devoid of legal effects and that national authorities/courts must take them into consideration when applying or interpreting binding rules (Case C-322/88, Court of Justice of the European Union 1989). At the same time, Expedia confirms that Commission notices do not necessarily bind national authorities or courts, even where they may provide relevant guidance (Case C-226/11, Court of Justice of the European Union 2012). This duty does not entail automatic conformity with soft instruments; rather, the authority is required to take relevant soft instruments into account through reasoned consideration such as weighing and addressing the soft-law arguments (Korkea-aho 2018, p. 474; Jiménez 2022, p. 10). In other words, legal effects as a mandatory interpretation aid do not dictate outcome, but require reasoned consideration in legal reasoning, whose intensity may vary according to the relevance, clarity and connection of the soft-law measure to the applicable EU framework (Liguori 2023).

Second, once the Commission codifies its assessment practice through guidelines, methodological frameworks, or similar instruments, EU administrative law may recognise self-binding effects. Therefore, such texts can constitute ‘rules of conduct’ that the Commission should follow. EU case-law recognises that where an institution adopts and publishes such rules to guide its discretion, it self-limits that discretion and cannot depart from them without giving reasons consistent with general principles of EU law, including equal treatment and legitimate expectations (Joined Cases C-189/02 P et al., Court of Justice of the European Union 2005). Kotnik confirms a related logic in the state-aid context: a Commission communication does not impose independent obligations on Member States, but it can structure the Commission’s own discretion and the conditions under which aid will be assessed (Case C-526/14, Court of Justice of the European Union 2016). Departures may require specific reasoning in order to avoid concerns under equality and legitimate expectations (Senden 2004, pp. 425–28). In this sense, legal effects are viewed as an indirect invalidating effect, because the operative constraint runs through hard-law obligations and general principles rather than through soft law as such (Senden 2004, pp. 425–28; Jiménez 2022, p. 17).

The operational questions for legal effects as a mandatory interpretative aid are: Is there an EU assessment point with consequences (e.g., a step in the enlargement process, a formal progress evaluation, or a conditionality decision)? Does the relevant EU document place the arrangements inside that assessment structure (as a criterion, benchmark, implementation item, or required deliverable), rather than confining them to a purely descriptive or political paragraph? For legal effects as indirect invalidating effects (self-binding), the questions are: Has the Commission codified a stable assessment practice (recurring headings, repeated benchmark items, standardised formulations across years/decisions)? Is the benchmark formulated with sufficient clarity and specificity to function as a “rule of conduct”, such that inconsistency can be framed in terms of general principles (not merely politics)? If so, would deviation normally require explicit justification to avoid concerns under equal treatment and/or legitimate expectations?

3. Content of Agreements

3.1. Content of the Brussels Agreement

The April 2013 Brussels Agreement is a 15-point attempt to institutionalise “normalisation” through governance integration rather than status settlement. Its architecture combines three elements. First, it provides for the creation of the ASM/CSM, framed through Kosovo law and the European Charter of Local Self-Government (Brussels Agreement 2013, points 1–6). The ASM/CSM is to be established by statute and endowed with structures (president, assembly, council) and a “full overview” in several policy sectors, while the text also anticipates “additional competences” delegated by the central level and a representative/monitoring role vis-à-vis central authorities (Brussels Agreement 2013, points 1–6). This combination is crucial: the Agreement offers a collective vehicle for Kosovo Serb municipal cooperation, but it does so through language that is sufficiently elastic to support competing expectations about the nature and authority of the ASM/CSM. Second, it seeks to integrate policing, judicial authority, and elections in northern Kosovo into Kosovo’s institutional and legal framework (Brussels Agreement 2013, points 7–12). The underlying aim is to dissolve dual authority in northern Kosovo by integrating core state functions: police are to be incorporated into the Kosovo Police under a regional commander for the four northern municipalities; the judiciary is to be integrated within the Kosovo legal framework, including an appellate panel and a permanent division sitting in Mitrovica; and municipal elections are to be organised with Organization for Security and Co-operation in Europe (OSCE) facilitation in accordance with Kosovo law (Brussels Agreement 2013, points 7–12). Third, the Agreement links implementation to the EU path through a mutual non-blocking clause and an EU-facilitated implementation committee, while energy and telecommunications are identified as spillover tracks (Brussels Agreement 2013, points 13–15).

In short, the Agreement operationalises a practice of integration without recognition and, therefore, of normalisation without settlement. Rather than resolving the status dispute, it substitutes institutional design for political consensus: northern Kosovo is integrated through police, judiciary, and elections organised within Kosovo’s legal framework, while the ASM/CSM provides a compensatory mechanism of collective municipal coordination that can be presented as compatible with non-recognition (Bieber 2015; Economides and Ker-Lindsay 2015). In this sense, the Brussels Agreement reflects the EU’s comparative advantage as mediator: it can structure bargaining, sequence issues, and apply accession-related incentives to produce incremental bargains even when core preferences on sovereignty remain unchanged (Bergmann 2018).

3.2. Content of the Ohrid Agreement

In the decade between the Brussels Agreement (2013) and the Ohrid Agreement (2023), the Dialogue was defined by a politics of non-implementation. The central trade-off of Kosovo Serbs’ accommodation within Pristina’s authority in return for an institutionalised form of Serb self-management (ASM/CSM) was never fully realised, mostly due to different interpretations of the ASM/CSM. The EU’s reliance on constructive ambiguity was useful for concluding agreements (under incompatible expectations), but it later enabled actors to weaponise competing interpretations, turning implementation into a struggle over meaning, authority, and constitutional identity (Beysoylu 2018; Troncotă 2018). While constructive ambiguity was not a sustainable solution on its own, the war in Ukraine reframed the Belgrade–Pristina normalisation through a more explicitly geopolitical lens and injected new urgency into the Dialogue (Kostovicova 2023; Vulović 2023; Đukanović and Vučić 2024).

The Ohrid Agreement establishes a status-neutral framework, proceeding “without prejudice” to the parties’ different views on the fundamental questions (European External Action Service [EEAS] 2023a, Preamble). While it avoids explicit reference to recognition, the parties commit to develop “normal, good-neighbourly relations” and to mutually recognise documents and national symbols (Art. 1), grounding this relationship in UN Charter principles including sovereign equality, territorial integrity, self-determination, human rights, and non-discrimination (Art. 2), alongside obligations of peaceful dispute settlement and the non-use of force (Art. 3). The Agreement further frames normalisation through four core commitments: neither party may represent the other internationally, Serbia will not object to Kosovo’s membership in any international organisation (Art. 4); neither side will block the other’s progress on the EU path (Art. 5), and both sides will continue negotiations under EU auspices, which “should lead to a legally binding agreement” (Art. 6). Next, Article 7 envisages an “appropriate level of self-management” for the Serbian community (including service provision, the possibility of Serbian financial support, and a direct communication channel to Kosovo’s government) in accordance with European experiences, and requires formalisation of the Serbian Orthodox Church’s status and protection of religious and cultural sites (Art. 7). Finally, the Agreement calls for the exchange of Permanent Missions (Art. 8), notes a prospective EU-led donor investment package (Art. 9), and establishes EU-chaired monitoring while reaffirming that all past Dialogue agreements remain valid and binding (Art. 10), with implementation guided by an annexed roadmap (Art. 11).

While the Ohrid Agreement sets out commitments, the Implementation Annex specifies implementation steps by tying them to EU leverage: benchmarks, monitoring, and funding consequences. It does so by defining itself as an integral part of the Agreement (European External Action Service [EEAS] 2023c, para. 1–2), and mandating an EU-chaired Joint Monitoring Committee (European External Action Service [EEAS] 2023c, para. 6). Most importantly, the Annex links the parties’ obligations to EU conditionality by incorporating them into Serbia’s Chapter 35 benchmarks and into the agenda of Kosovo’s Special Group on Normalisation (European External Action Service [EEAS] 2023c, para. 3), prioritising endorsement of the Declaration on Missing Persons (European External Action Service [EEAS] 2023c, para. 4) and the immediate launch of negotiations on the Article 7 arrangements in line with previous agreements (European External Action Service [EEAS] 2023c, para. 5). Financially, the Annex links the promised donor package to conditionality since no disbursement will occur until the EU determines that all provisions have been fully implemented (European External Action Service [EEAS] 2023c, para. 7). Finally, it clarifies implementation rules: all articles are to be implemented independently and without prejudice as to sequencing (European External Action Service [EEAS] 2023c, para. 8–9), commits both sides not to block implementation (European External Action Service [EEAS] 2023c, para. 10–11), and explicitly warns that failures may trigger direct negative consequences for enlargement progress and EU financial assistance (European External Action Service [EEAS] 2023c, para. 12).

The Ohrid Agreement and its Annex can therefore be read as the EU’s attempt to reset the Dialogue by moving towards a basic treaty–style framework, in the sense of a structured package of reciprocal normalisation commitments without formal recognition, while tying implementation more explicitly to enlargement benchmarks and EU financial leverage (Istrefi 2023; Kreft 2023; Đukanović and Vučić 2024).

4. Discussion

This section applies Tests A–C to the Brussels and Ohrid Agreements. The aim is not to force a binary label, but to identify which bindingness claims are supported by the text and subsequent practice, and which are not.

4.1. Test A: Animus Contrahendi

4.1.1. Brussels Agreement

(a) Manifestation. The Brussels Agreement is unusually concrete for a purportedly political arrangement. It establishes an ASM/CSM with defined structures and a “full overview” in specified sectors; provides for the integration of police and judiciary within Kosovo’s framework; foresees OSCE-facilitated local elections in accordance with Kosovo law; and includes an EU-facilitated implementation modality alongside a mutual non-blocking clause linked to the parties’ EU paths.

(b) Intent indicators. The combination of institutional design, implementation sequencing, and an EU-chaired follow-up mechanism points beyond aspirational language. The agreement is written in a register of institutional commitments (what is “to be” created or integrated) and embeds the bargain in procedures that presuppose acceptance of an agreed allocation of authority. At the same time, deliberate elasticity (especially around the nature and authority of the ASM/CSM) complicates the inference of a shared, stable intent to be legally bound in a determinate way. The strongest treaty-like signal comes from Kosovo’s internal treatment of the Brussels Agreement as an international agreement: the Assembly adopted Law No. 04/L-199 on ratification (Assembly of the Republic of Kosovo 2013). This does not by itself settle inter se treaty status, but it materially strengthens the inference that at least one party (at one point) publicly accepted the instrument as normatively operative. On the Serbian side, while there was no classic treaty ratification, the National Assembly adopted the Government’s report on the negotiations with Pristina by a large majority (Narodna skupština Republike Srbije 2013), signalling institutional authorisation and political commitment, albeit in a weaker (and different) register.

(c) Reliance/performance. The Brussels Agreement clearly aims to operationalise “integration without recognition” by substituting institutional design for political consensus. Subsequent practice, however, has been characterised by contestation and selective implementation, especially around the central trade-off of the ASM/CSM. As a result, implementation became a struggle over meaning and authority. This pattern is not itself decisive evidence against intent,2 but it does suggest that the very ambiguity used to produce the text created difficulties in sustaining shared understanding of obligations. This fragility is not merely political: the Kosovo Constitutional Court held that the later “General Principles/Main Elements” for the ASM/CSM were “not entirely in compliance” with the spirit of the Constitution and several constitutional provisions. This legally constrained the translation of Brussels commitments into domestic form (Constitutional Court of the Republic of Kosovo 2015, pp. 37–38). Taken together, the Brussels Agreement shows high manifestation but mixed and partly asymmetric intent signals, while subsequent practice reveals persistent contestation over the instrument’s operative meaning; thus, on Test A alone, it falls short of a demonstrable treaty-type obligation.

4.1.2. Ohrid Agreement

Applying Test A to the Ohrid Agreement requires balancing two features that pull in opposite directions: its distinctly treaty-like drafting, while its internal language can be read as pointing towards a future legally binding instrument. (a) On manifestation, the text is drafted as an instrument of obligations rather than a set of minutes: it is structured in numbered Articles (1–11) and contains clear reciprocal commitments. It goes beyond the language of “good-neighbourly relations” and includes a set of “will not” obligations such as not blocking the other party’s participation in international organisations and not representing the other in the international sphere. The text also contains implementation-relevant commitments (e.g., mutual protection of rights, exchanges of permanent missions, dispute-settlement expectations), which further strengthens the impression of a self-contained obligations package rather than a mere declaration.

(b) The difficulty, however, arises at the level of intent. Article 6 states that the parties “shall continue their efforts” and that “negotiations shall conclude with an agreement… which should lead to a legally binding agreement”. This formulation can be read as indicating that the text’s purpose is to pave the way for bindingness later, rather than itself embodying animus contrahendi. The politics of signature reinforce that ambiguity. Immediately after the meeting, Serbia’s President publicly stated that Serbia would work towards normalisation with Kosovo but would not “sign any international legally binding documents with Kosovo because Serbia does not recognise its independence” (Reuters 2023), whereas the EU has reiterated that the Agreement and its Implementation Annex are binding in their entirety under international law despite the absence of signatures.

(c) Reliance/performance. The content of the Ohrid Agreement supplies stronger treaty-like signals than the Brussels Agreement. However, given its recency, subsequent practice remains limited as an indicator for animus contrahendi. The most defensible conclusion is that the Ohrid Agreement is closer to a treaty analogue in structure and obligation-language, but Article 6 and the refusal to sign complicate any categorical claim that it reflects a settled shared intent to be bound inter se. Its bindingness against Test A therefore remains contestable and will depend on how Article 6 is interpreted in future practice and the Annex’s operationalisation mechanisms.

4.2. Test B: Subsequent Conduct and Acquiescence

4.2.1. Brussels Agreement

(a) Manifestation. The Brussels Agreement generates a recognisable set of practices that are public, attributable, and (at least initially) mutually engaged. The clearest authority-allocating practice object is the integration of Serbian structures into Kosovo’s institutional framework. In practice, the integration of the police officers was completed in 2014 (European Commission 2014, p. 4), while Kosovo Serb judges and prosecutors took the oath and were integrated into Kosovo’s judiciary in 2017 (European Commission 2022, p. 18; see also European Western Balkans 2017). These were not isolated gestures: they were visible, operational acts that necessarily called for reaction and coordination by the other side. Simultaneously, the first Kosovo-wide elections were held in late 2013 and “municipalities in the north of Kosovo were inaugurated in conformity with Kosovo law” (European Commission 2014, p. 5), providing an additional, repeatable practice object under the assumed allocation of authority.

(b) Intent indicators. The subsequent record shows mixed signals. While early implementation can be read as the operation of routine procedures on the assumed allocation of authority, later developments introduce explicit public objections that complicate any inference of settled acquiescence. The most notable event is the mass resignations of Kosovo Serb officials in November 2022 (European Commission 2025b, p. 61), justified on the ground that Pristina “constantly violates international public law and the Brussels Agreement”, and that Kosovo Serbs are exposed to “unceasing and continuous terror and all kinds of violence” by Pristina’s institutions (European Western Balkans 2022). The resignations included mayors, MPs, police officers, and judicial personnel, and they were followed by an election boycott within the Kosovo Serb community (European Commission 2024, p. 4). Kosovo Serb participation also proved reversible: Serb representatives returned to municipal office after the October 2025 local elections (Bami 2025; European Commission 2025b, p. 61). However, this political re-entry did not automatically translate into full institutional return across the system (European Commission 2025b, p. 63).

(c) Reliance and performance: Those behaviours on both sides evolved in ways that are difficult to explain without at least a degree of operational acceptance—above all the institutional integration steps and election practices that embed assumptions about authority in day-to-day governance. However, that operational acceptance is uneven and politically reversible across practice objects, and the most contested elements (notably the non-establishment of the ASM/CSM) do not sustain an inference of convergent, durable acquiescence sufficient to support bindingness for the arrangement as a whole.

4.2.2. Ohrid Agreement

(a) Manifestation. Even though it is a relatively recent document, the Ohrid Agreement produces several observable and attributable practices. The most explicit institutional practice object is the Joint Monitoring Committee (JMC) (Art. 10), which was formally established in April 2023 under EU chairmanship (European External Action Service [EEAS] 2023b). The missing-persons track also offers a clear sequence of acts: thus, the parties adopted the Declaration (European External Action Service [EEAS] 2023d), agreed on the terms of reference for a Joint Commission (European Commission 2025c, p. 74), and held the first trilateral Joint Commission meeting in January 2026 (European External Action Service [EEAS] 2026). Article 4 (“Serbia will not object to Kosovo’s membership in any international organisation”), by its very nature, creates numerous opportunities for observable objections (or lack of objections). The documents/symbols cluster (Art. 1) manifests most concretely through the licence-plates arrangement, which generated visible administrative implementation in late 2023/early 2024 (European Commission 2025c, p. 73).

(b) Intent indicators. The intent signal is mixed. While the establishment of the JMC is consistent with procedural assent, subsequent implementation points more to episodic engagement than to settled acceptance, since it has not resulted in regular meetings or a published work plan (European Institute of Peace 2025, p. 17). The Joint Commission (for missing persons) shows the strongest movement towards institutionalisation, given the adoption of the Declaration, the establishment of the Commission, and the agreement on its Terms of Reference. The main negative signal is the delay between these concrete outcomes. Article 4 moves in the opposite direction as Serbia continues to lobby against Kosovo’s membership in international organisations, which points towards a clear objection (Beširević 2025; European Commission 2025c, p. 72). As for the documents/symbols cluster, licence plates suggest operational acceptance, since Serbia stopped issuing licence plates with Kosovo city denominations and the parties moved towards “the mutual recognition of licence plates”; however, other elements remain too incomplete to support a broad inference of acceptance (European Commission 2025c, pp. 71–73).

(c) Reliance/performance. Reliance and performance are strongest in the case where routines become costly to unwind, such as licence plates: once implemented, the rule embeds itself in day-to-day border/administrative practice and produces predictable reliance. By contrast, the JMC does not (yet) function as a stable routine, and Article 4 is an example of contrary-performance where there is a clear objection. The Joint Commission lies in between, as there are some working institutional mechanisms, but it falls short of durable practice.

4.3. Test C: EU Legal Effects

Test C is applied in two steps. First, it asks whether the agreements have been incorporated into EU assessment structures that carry legal or administrative consequences, such as accession benchmarks, progress reporting, monitoring mechanisms or funding conditionality. Second, it asks whether the Commission and other EU institutions have treated implementation as a recurring benchmark, formulated with sufficient clarity and consistency to require reasoned consideration or to structure institutional discretion. The question is therefore not whether the Brussels or Ohrid Agreements become binding agreements as such, but whether they operate as legally relevant reference points within the EU’s enlargement and conditionality framework.

4.3.1. Brussels Agreement

The key document for legal effects as a mandatory interpretative aid in relation to the Brussels Agreement is the EU Common Position on Chapter 35 (30 November 2015), an accession-negotiation document adopted within Serbia’s EU accession framework rather than an ordinary secondary-law instrument. The document explicitly ties normalisation between Belgrade and Pristina to Serbia’s progress on the EU path by placing it within Chapter 35 and by measuring the advancement of negotiations against Serbia’s continued engagement towards a visible and sustainable improvement in relations with Kosovo. Most importantly, it creates an assessment point with consequences: if progress in normalisation “significantly lags behind” overall negotiation progress due to Serbia failing to act in good faith—particularly in implementing agreements—the Commission will, on its own initiative or at the request of one third of the Member States, “propose to withhold its recommendations to open and/or close other negotiating chapters… until this imbalance is addressed” (Council of the European Union 2015, p. 3). The document also establishes continuous monitoring and at least twice-yearly reporting to the Council on interim benchmarks. Therefore, the EU cannot treat the Brussels Agreement as legally irrelevant within the accession framework: once Brussels-linked implementation is embedded in a consequence-bearing assessment structure, it must be taken into account in reasoned EU assessment (a duty of consideration rather than automatic conformity).

Besides this mandatory interpretative aid effect, the Commission’s reporting practice can support an indirect invalidating (self-binding) effect. The annual reports use standardised compliance benchmarks: rather than merely describing political developments, they evaluate conduct against a stable set of Dialogue “obligations” and past agreements. In the Serbia report, the Commission characterises the withdrawal of Kosovo Serbs from institutions and the boycott of local elections as violations of Serbia’s Dialogue obligations and as “severe backsliding” in compliance with the April 2013 Brussels Agreement. It also describes these developments as involving a “direct violation” of the 2015 Justice Agreement and adds that “both Parties are not complying” with those agreements at the time of reporting (European Commission 2025c, p. 72). It then sets out the benchmarked pathway for reversal, recalling that the return of officials “has to be done in accordance with past Dialogue agreements and with full respect for Kosovo’s legal framework” (European Commission 2025c, p. 72). On this basis, the legal effect is not that the Brussels text becomes binding as such, but that the Commission’s codified benchmark script can operate as a “rule of conduct” in the exercise of its hard-law competences in enlargement assessment and conditionality. Once the Commission repeatedly structures its reasoning around the same reference points—Dialogue obligations; compliance/non-compliance; specified agreements; conditions for “return”—those reference points become an evaluative baseline. A departure from that evaluative baseline in comparable cases would ordinarily require explicit justification to avoid concerns under general principles of EU law, such as consistency/equal treatment (and, where reliance could be shown, legitimate expectations). In this sense, the effect is indirect invalidating: the operative constraint runs through general principles governing EU administrative action, rather than through any direct binding force of the Brussels arrangements.

4.3.2. Ohrid Agreement

The Implementation Annex (European External Action Service [EEAS] 2023c) is explicitly designed to create a path for (at least) legal effects as a mandatory interpretative aid. It presents the Agreement and Annex as integral to the parties’ accession processes, provides for EU-chaired monitoring, and envisages that the EU Facilitator will start the process to amend Serbia’s Chapter 35 benchmarks and that Kosovo’s Special Group on Normalisation will reflect the new obligations in its agenda. The Annex also adds a conditionality layer: it ties the promised donor package to an EU determination of full implementation and warns that failures can trigger negative consequences for enlargement progress and EU financial assistance. Beyond the Annex itself, the Reform and Growth Facility for the Western Balkans establishes a binding precondition for financial support requiring measurable progress and tangible results, including full implementation of obligations stemming from the Ohrid Agreement and its Annex (European Union 2024, Art. 5(1)). The practical limits of this conditionality are visible in the subsequent implementation record: the envisaged donor conference was postponed due to the absence of progress, while the EU’s attention increasingly shifted from implementation of the Agreement to crisis management and de-escalation in northern Kosovo (Mirosavljević and Burazer 2025). In this setting, the EU must determine whether implementation has occurred in order to decide on disbursement and progress, which structurally requires taking the benchmark into account. This places the Ohrid Agreement within a formal EU assessment structure in which progress/implementation is treated as a criterion for evaluating the normalisation track. Therefore, even if the Ohrid Agreement’s international-law bindingness is disputed, EU authorities cannot regard it as having no legal effect when applying their own funding and assessment instruments; at minimum, it must be addressed as an operative benchmark in EU reasoning.

Additionally, there is evidence that the Commission has begun to codify a stable assessment practice—a repeatable “rule of conduct” in EU evaluative reasoning—such that deviation would ordinarily require explicit justification under general principles, supporting an indirect invalidating (self-binding) effect. The 2025 Enlargement Policy Communication reiterates, in standardised terms, that “both Kosovo and Serbia are expected to implement all of their respective obligations stemming from the Agreement on the Path to Normalisation and its Implementation Annex, as well as all past Dialogue agreements”, and it links progression on the EU path to “meaningful progress in the normalisation of their relations” (European Commission 2025a, p. 22). It further warns that, absent constructive engagement, “both parties risk missing major opportunities” and expressly notes that moving forward on normalisation is also “a requirement under the Reform and Growth Facility for the Western Balkans” (European Commission 2025a, p. 22). The Kosovo and Serbia reports mirror this evaluative script by framing normalisation in benchmarked form and treating implementation/progress as a recurring criterion within the annual assessment architecture (European Commission 2025b, pp. 63–64; European Commission 2025c, pp. 71–73). As with the Brussels Agreement, the effect is not that the Ohrid Agreement is binding as such, but that the Commission has publicly and consistently codified the benchmark logic across its enlargement instruments; consequently, a sudden departure from that evaluative baseline without reasons can, in principle, be framed as inconsistency in the exercise of hard-law powers, with the operative constraint running through general principles rather than through soft law “as such”.

5. Conclusions

This article set out to clarify a persistent source of confusion in debates on the Belgrade–Pristina Dialogue: the tendency to treat “legally binding” as a single label, rather than two analytically distinct claims. By separating binding force under international law from legal effects produced through the EU legal order, the article explains how the same Agreements can remain contestable as treaty-type obligations while still functioning as operative constraints within enlargement and conditionality practice. Methodologically, it drew on ICJ indicia for intent and acceptance outside ordinary treaty form (Tests A and B), and then examined how EU soft-law governance can generate legally relevant effects through consequence-bearing assessment points and institutional self-binding (Test C).

On the question of international-law binding force, the analysis suggests a cautious, differentiated conclusion. The Brussels Agreement displays unusually high manifestation for a purportedly political understanding and was followed by significant early implementation in core governance domains; yet the combination of deliberate textual elasticity, asymmetric domestic treatment, and subsequent contestation undermines a confident inference of shared animus contrahendi sufficient to sustain treaty-type obligation. The Ohrid Agreement, by contrast, is drafted in a more treaty-like register and contains clearer reciprocal commitments, but Article 6 and the politics of non-signature complicate any categorical claim that the parties intended to create legal relations inter se at the moment of adoption. Given the instrument’s recency and mixed subsequent practice across practice objects, the most defensible position is that the Ohrid Agreement is closer to a treaty analogue in form, yet remains contestable in legal character until sustained practice and interpretative handling of Article 6 consolidate (or undermine) a shared intent to be bound.

The more robust finding of the article lies in the EU-law dimension. Whatever the limits of treaty-type bindingness, the Brussels and Ohrid Agreements are embedded in EU instruments that create assessment points with consequences—opening/closing dynamics, progress evaluations, and, increasingly, funding decisions conditioned on measurable implementation. Once these Agreements are operationalised as benchmarks within consequence-bearing EU decision structures, EU authorities cannot treat them as having “no legal effect”: at minimum, they become mandatory reference points that must be taken into account in reasoned assessment; and where the Commission codifies a stable evaluative script across years and decisions, that practice may generate self-binding effects whose constraint runs through general principles governing EU administrative action rather than through the agreements “as such”.

This has three implications. First, the practical “bindingness” of the Dialogue is driven less by formal treaty status than by the EU’s capacity to translate political arrangements into conditionality benchmarks with consequences. Second, for the parties, contesting whether an arrangement is a treaty does not, by itself, neutralise EU leverage where implementation is made legally relevant to accession and financing decisions: the operative pressure point is the EU’s own decision-making architecture. Third, for scholarship, treating the Dialogue instruments as simply “hard” or “soft” law risks missing the core mechanism at work—hybrid governance in which political bargains can remain legally contestable under international law while becoming legally consequential through EU institutional practice.

Two caveats follow. The binding-force assessment is inherently time-sensitive: sustained, concordant practice; stable institutional routines that are costly to unwind; and a consistent pattern of non-objection could strengthen an inference of acceptance over time, particularly with respect to the Ohrid Agreement’s more obligation-like drafting. Conversely, persistent contrary-performance and explicit contestation may continue to limit any claim of shared intent. Future work should therefore track how EU benchmark codification evolves across successive enlargement cycles and funding decisions, and how the parties’ conduct—especially around the most contested obligations—either consolidates or fragments the understanding of what, if anything, is accepted as obligatory.

In short, the Brussels and Ohrid Agreements are weak candidates for classic treaty-type binding force, but strong candidates for EU-generated legal effects. The Dialogue’s “bindingness” is therefore best understood as conditionality-driven operability rather than consent-based obligation in the traditional treaty sense.