Non-invasive respiratory support (NIRS) is widely used in COVID-19 patients, although high rates of NIRS failure are reported. Early detection of NIRS failure and promptly defining the need for intubation are crucial for the management of patients with acute respiratory failure (ARF). We tested the ability of the HACOR score¸ a scale based on clinical and laboratory parameters, to predict adverse outcomes in hospitalized COVID-19 patients with ARF. Four hundred patients were categorized according to high (>5) or low (≤5) HACOR scores measured at baseline and 1 h after the start of NIRS treatment. The association between a high HACOR score and either in-hospital death or the need for intubation was evaluated. NIRS was employed in 161 patients. Forty patients (10%) underwent intubation and 98 (25%) patients died. A baseline HACOR score > 5 was associated with the need for intubation or in-hospital death in the whole population (HR 4.3; p < 0.001), in the subgroup of patients who underwent NIRS (HR 5.2; p < 0.001) and in no-NIRS subgroup (HR 7.9; p < 0.001). In the NIRS subgroup, along with the baseline HACOR score, also 1-h HACOR score predicted NIRS failure (HR 2.6; p = 0.039). In conclusion, the HACOR score is a significant predictor of adverse clinical outcomes in patients with COVID-19-related ARF.