BACKGROUND: The use of high-flow nasal cannula (HFNC) and noninvasive ventilation (NIV) for hypoxemic respiratory failure secondary to COVID-19 are recommended by critical-care guidelines; however, apprehension about viral particle aerosolization and patient self-inflicted lung injury may have limited use. We aimed to describe hospital variation in the use and clinical outcomes of HFNC and NIV for the management of COVID-19. METHODS: This was a retrospective observational study of adults hospitalized with COVID-19 who received supplemental oxygen between February 15, 2020, and April 12, 2021, across 102 international and United States hospitals by using the COVID-19 Registry. Associations of HFNC and NIV use with clinical outcomes were evaluated by using multivariable adjusted hierarchical random-effects logistic regression models. Hospital variation was characterized by using intraclass correlation and the median odds ratio. RESULTS: Among 13,454 adults with COVID-19 who received supplemental oxygen, 8,143 (60%) received nasal cannula/face mask only, 2,859 (21%) received HFNC, 878 (7%) received NIV, 1,574 (12%) received both HFNC and NIV, with 3,640 subjects (27%) progressing to invasive ventilation. The hospital of admission contributed to 24% of the risk-adjusted variation in HFNC and 30% of the risk-adjusted variation in NIV. The median odds ratio for hospital variation of HFNC was 2.6 (95% CI 1.4–4.9) and of NIV was 3.1 (95% CI 1.2-8.1). Among 5,311 subjects who received HFNC and/or NIV, 2,772 (52%) did not receive invasive ventilation and survived to hospital discharge. Hospital-level use of HFNC or NIV were not associated with the rates of invasive ventilation or mortality. CONCLUSIONS: Hospital variation in the use of HFNC and NIV for acute respiratory failure secondary to COVID-19 was great but was not associated with intubation or mortality. The wide variation and relatively low use of HFNC/NIV observed within our study signaled that implementation of increased HFNC/ NIV use in patients with COVID-19 will require changes to current care delivery practices. (ClinicalTrials.gov registration NCT04323787.).