Objectives High flow nasal cannula (HFNC) is frequently used in patients with acute respiratory failure, but there is limited evidence regarding predictors of therapeutic failure. The objective of this study was to assess diaphragmatic ultrasound criteria as predictors of failure to HFNC, defined as the need for orotracheal intubation or death. Methods Prospective cohort study including adult patients consecutively admitted to the critical care unit, from July 24 to October 20, 2020, with respiratory failure secondary to SARS‐CoV‐2 pneumonia who required HFNC. After 12 hours of HFNC initiation we measured ROX index (ratio of SpO 2 /FiO 2 to respiratory rate), excursion and diaphragmatic contraction speed (diaphragmatic excursion/inspiratory time) by ultrasound, both in supine and prone position. Results In total, 41 patients were analyzed, 25 succeeded and 16 failed HFNC therapy. At 12 hours, patients who succeeded HFNC therapy presented higher ROX index in supine position (9.8 [9.1–15.6] versus 5.4 [3.9–6.8], P < .01), and higher PaO 2 /FiO 2 ratio (186 [135–236] versus 117 [103–162] mmHg, P = .03). To predict therapeutic failure, the supine diaphragmatic contraction speed presented sensitivity of 89% and a specificity of 57%, while the ROX index presented a sensitivity of 92.8% and a specificity of 75%. Conclusions Diaphragmatic contraction speed by ultrasound emerges as a diagnostic complement to clinical tools to predict HFNC success. Future studies should confirm these results.