Objective: We aim to determine the impact of steroid use in COVID-19 pneumonia in-hospital mortality. Design: We performed a single-centre retrospective cohort study. Setting: A University hospital in Madrid, Spain, during March 2020. Participants: Patients admitted with SARS-CoV-2 pneumonia. Exposures: Patients treated with steroids were compared to patients not treated with steroids. A propensity-score for steroid treatment was developed. Different steroid regimens were also compared, and adjusted with a second propensity score. Main Outcomes and Measures: To determine the role of steroids in in-hospital mortality, univariable and multivariable analyses were performed, and adjusted including the propensity score as a covariate. Survival times were compared using a log-rank test. Results: During the study period, 463 out of 848 hospitalized patients with COVID19 pneumonia fulfilled inclusion criteria. Among them, 396 (46.7%) consecutive patients were treated with steroids and 67 patients were assigned to the control cohort. Global mortality was 15.1%. Median time to steroid treatment from symptom onset was 10 days (IQR 8 to13). In-hospital mortality was lower in patients treated with steroids than in controls (13.9% [55/396] versus 23.9% [16/67], OR 0.51 [0.27 to 0.96], p= 0.044). Steroid treatment reduced mortality by 41.8% relative to no steroid treatment (RRR 0,42 [0.048 to 0.65). Initial treatment with 1 mg/kg/day of methylprednisolone (or equivalent) versus steroid pulses was not associated with in-hospital mortality (13.5% [42/310] versus 15.1% [13/86], OR 0.880 [0.449-1.726], p=0.710). Conclusions: Our results show that survival of patients with SARS-CoV2 pneumonia is higher in patients treated with glucocorticoids than in those not treated. In-hospital mortality was not different between initial regimens of 1 mg/kg/day of methylprednisolone or equivalent and glucocorticoid pulses. These results support the use of glucocorticoids in SARS-CoV2 infection.