INTRODUCTION: High-flow nasal cannula (HFNC) therapy in patients with hypoxemic respiratory failure due to COVID-19 is poorly understood and remains controversial. METHODS: We evaluated a large cohort of patients with COVID-19-related hypoxemic respiratory failure at the temporary COVID-19 hospital in Mexico City. The primary outcome was the success rate of HFNC to prevent the progression to invasive mechanical ventilation (IMV). We also evaluated the risk factors associated with HFNC success or failure. RESULTS: HFNC use effectively prevented IMV in 71.4% of patients [270 of 378 patients; 95% confidence interval (CI) 66.6-75.8%]. Factors that were significantly different at admission included age, the presence of hypertension, and the Charlson comorbidity index. Predictors of therapy failure (adjusted hazard ratio, 95% CI) included the comorbidity-age-lymphocyte count-lactate dehydrogenase (CALL) score at admission (1.27, 1.09-1.47; p < 0.01), Rox index at 1 hour (0.82, 0.7-0.96; p = 0.02), and no prior steroid treatment (0.34, 95% CI 0.19-0.62; p < 0.0001). Patients with HFNC success rarely required admission to the intensive care unit and had shorter lengths of hospital stay [19/270 (7.0%) and 15.0 (interquartile range, 11-20) days, respectively] than those who required IMV [104/108 (96.3%) and 26.5 (20-36) days, respectively]. CONCLUSION: Treating patients with HFNC at admission led to improvement in respiratory parameters in many patients with COVID-19.