Background: Convalescent plasma is a potential therapeutic option for critically ill patients with coronavirus disease 19 (COVID-19), yet its efficacy remains to be determined . The aim was to investigate the effects of convalescent plasma (CP) in critically ill patients with COVID-19.Methods: This was a single-center prospective observational study conducted in Rio de Janeiro, Brazil, from March 17th to May 30th, with final follow-up on June 30th . We included 113 laboratory-confirmed COVID-19 patients with respiratory failure. Primary outcomes were time to clinical improvement and survival within 28 days . Secondary outcomes included behavior of biomarkers and viral loads . Kaplan–Meier analyses and Cox proportional-hazards regression using propensity score with inverse-probability weighing were performed.Results : 41 patients received CP and 72 received standard of care (SOC). Median age was 61 years (IQR 48–68), disease duration was 10 days (IQR 6–13), and 86% were mechanically ventilated . At least 29 out of 41CP-recipients had baseline IgG titers ≥ 1:1,080 . Clinical improvement within 28 days occurred in 19 (46 %) CP-treated patients, as compared to 23 (32 %) in the SOC group [adjusted hazard ratio (aHR) 0.91 (0.49–1.69) ]. There was no significant change in 28-day mortality (CP 49% vs. SOC 56% ; aHR 0.90 [0.52–1.57] ). Biomarker assessment revealed reduced inflammatory activity and increased lymphocyte count after CP.Conclusions: In this study, CP was not associated with clinical improvement or increase in 28-day survival . However, our study may have been underpowered and included patients with high IgG titers and life-threatening disease.Clinical Trial Registration: The study protocol was retrospectively registered at the Brazilian Registry of Clinical Trials (ReBEC) with the identification RBR-4vm3yy (http: //www.ensaiosclinicos.gov.br).