Understanding the cause of sex disparities in COVID-19 outcomes is a major challenge. We investigate sex hormone levels and their association with outcomes in COVID-19 patients, stratified by sex and age. This observational, retrospective, cohort study included 138 patients aged 18 years or older with COVID-19, hospitalized in Italy between February 1 and May 30, 2020. The association between sex hormones (testosterone, estradiol, progesterone, dehydroepiandrosterone) and outcomes (ARDS, severe COVID-19, in-hospital mortality) was explored in 120 patients aged 50 years and over. STROBE checklist was followed. The median age was 73.5 years IQR 61, 82; 55.8% were male. In older males, testosterone was lower if ARDS and severe COVID-19 were reported than if not (3.6 5.3 nmol/L, p =0.0378 and 3.7 8.5 nmol/L, p =0.0011, respectively). Deceased males had lower testosterone (2.4 4.8 nmol/L, p =0.0536) and higher estradiol than survivors (40 24 pg/mL, p = 0.0006). Testosterone was negatively associated with ARDS (OR 0.849 95% CI 0.734, 0.982), severe COVID-19 (OR 0.691 95% CI 0.546, 0.874), and in-hospital mortality (OR 0.742 95% CI 0.566, 0.972), regardless of potential confounders, though confirmed only in the regression model on males. Higher estradiol was associated with a higher probability of death (OR 1.051 95% CI 1.018, 1.084), confirmed in both sex models. In males, higher testosterone seems to be protective against any considered outcome. Higher estradiol was associated with a higher probability of death in both sexes.