Background: Coronavirus-associated acute respiratory distress syndrome (CARDS) has limited effective therapy to date. NLRP3 inflammasome activation induced by SARS-CoV-2 in COVID-19 contributes to cytokine storm. Methods: This randomised, multinational study enrolled hospitalised patients (18–80 years) with COVID-19-associated pneumonia and impaired respiratory function. Eligible patients were randomised (1:1) via Interactive Response Technology to DFV890 + standard-of-care (SoC) or SoC alone for 14 days. Primary endpoint was APACHE II score at Day 14 or on day-of-discharge (whichever-came-first) with worst-case imputation for death. Other key assessments included clinical status, CRP levels, SARS-CoV-2 detection, other inflammatory markers, in-hospital outcomes, and safety. Findings: Between May 27, 2020 and December 24, 2020, 143 patients (31 clinical sites, 12 countries) were randomly assigned to DFV890 + SoC (n = 71) or SoC alone (n = 72). Primary endpoint to establish clinical efficacy of DFV890 vs. SoC, based on combined APACHE II score, was not met; LSM (SE), 8·7 (1.06) vs. 8·6 (1.05); p = 0.467. More patients treated with DFV890 vs. SoC showed ≥ 1-level improvement in clinical status (84.3% vs. 73.6% at Day 14), earlier clearance of SARS-CoV-2 (76.4% vs. 57.4% at Day 7), and mechanical ventilation-free survival (85.7% vs. 80.6% through Day 28), and there were fewer fatal events in DFV890 group (8.6% vs. 11.1% through Day 28). DFV890 was well tolerated with no unexpected safety signals. Interpretation: DFV890 did not meet statistical significance for superiority vs. SoC in primary endpoint of combined APACHE II score at Day 14. However, early SARS-CoV-2 clearance, improved clinical status and in-hospital outcomes, and fewer fatal events occurred with DFV890 vs. SoC, and it may be considered as a protective therapy for CARDS. Trial registration: ClinicalTrials.gov, NCT04382053.