Purpose: Tocilizumab is associated with positive outcomes in severe COVID-19. We wanted to describe the characteristics of nonresponders to treatment. Methods: This was a retrospective multicenter study in two respiratory departments investigating adverse outcomes at 90 days from diagnosis in subjects treated with tocilizumab (8 mg/kg intravenously single dose) for severe progressive COVID-19. Results: Of 121 subjects, 62% were males, and 9% were fully vaccinated. Ninety-six (79.4%) survived, and 25 died (20.6%). Compared to survivors (S), nonsurvivors (NS) were older (median 57 versus 75 years of age), had more comorbidities (Charlson comorbidity index 2 versus 5) and had higher rates of intubation/mechanical ventilation (p
< 0.05). On admission, NS had a lower PO2
ratio, higher blood ferritin, and higher troponin, and on clinical progression (day of tocilizumab treatment), NS had a lower PO2
ratio, decreased lymphocytes, increased neutrophil to lymphocyte ratio, increased ferritin and lactate dehydrogenase (LDH), disease located centrally on computed tomography scan, and increased late c-reactive protein. Cox proportional hazards regression analysis identified age and LDH on deterioration as predictors of death; admission PO2
ratio and LDH as predictors of intubation; PO2
ratios, LDH, and central lung disease on radiology as predictors of noninvasive ventilation (NIV) (a < 0.05). The log-rank test of mortality yielded the same results (p
< 0.001). ROC analysis of the above predictors in a separate validation cohort yielded significant results. Conclusions: Older age and high serum LDH levels are predictors of mortality in tocilizumab-treated severe COVID-19 patients. Hypoxia levels, LDH, and central pulmonary involvement radiologically are associated with intubation and NIV.