Importance: Serum ferritin, an acute phase marker of inflammation, has several physiologic functions, including limiting intracellular oxidative stress. Whether the effectiveness of corticosteroids differs according to serum ferritin level in COVID-19 has not been reported. Objective: To examine the association between admission serum ferritin level and methylprednisolone treatment outcomes in nonintubated patients with severe COVID-19. Design, Setting, and Participants: This retrospective cohort study included patients with severe COVID-19 admitted to an academic referral center in Stony Brook, New York, from March 1 to April 15, 2020, receiving high-flow oxygen therapy (fraction of inspired oxygen, ≥50%). The outcomes of treatment with methylprednisolone were estimated using inverse probability of treatment weights, based on a propensity score comprised of clinical and laboratory variables. Patients were followed up for 28 days. Data were analyzed from December 19, 2020, to July 22, 2021. Exposures: Systemic methylprednisolone administered per the discretion of the treating physician. Main Outcomes and Measures: The primary outcome was mortality, and the secondary outcome was a composite of death or mechanical ventilation at 28 days. Results: Among 380 patients with available ferritin data (median [IQR] age, 60 years [49-72] years; 130 [34.2%] women; 250 [65.8%] men; 310 White patients [81.6%]; 47 Black patients [12.4%]; 23 Asian patients [6.1%]), 142 patients (37.4%) received methylprednisolone (median [IQR] daily dose, 160 [120-240] mg). Ferritin levels were similar in patients who received methylprednisolone vs those who did not (median [IQR], 992 [509-1610] ng/mL vs 893 [474-1467] ng/mL; P =.32). In weighted analyses using tertiles of ferritin values (lower: 29-619 ng/mL; middle: 623-1316 ng/mL; upper: 1322-13418 ng/mL), methylprednisolone was associated with lower mortality in patients with ferritin in the upper tertile (HR, 0.16; 95% CI, 0.06-0.45) and higher mortality in those with ferritin in the middle (HR, 2.46; 95% CI, 1.15-5.28) and lower (HR, 2.43; 95% CI, 1.13-5.22) tertiles (P for interaction <.001). Composite end point rates were lower with methylprednisolone in patients with ferritin in the upper tertile (HR, 0.45; 95% CI, 0.25-0.80) but not in those with ferritin in the middle (HR, 0.83; 95% CI, 0.50-1.39) and lower (HR, 0.89; 95% CI, 0.51-1.55) tertiles (P for interaction =.11). Conclusions and Relevance: In this cohort study of nonintubated patients with severe COVID-19, methylprednisolone was associated with improved clinical outcomes only among patients with admission ferritin in the upper tertile of values..