BACKGROUND Mortality among patients admitted to intensive care units (ICUs) with COVID-19 is unclear due to variable follow-up periods. Few nationwide data are available to compare risk factors, treatment and outcomes of COVID-19 patients after ICU admission. OBJECTIVE To evaluate baseline characteristics, treatments and 30-day outcomes of patients admitted to Swedish ICUs with COVID-19. DESIGN Registry-based cohort study with prospective data collection. SETTING Admissions to Swedish ICUs from 6 March to 6 May 2020 with laboratory confirmed COVID-19 disease. PARTICIPANTS Adult patients admitted to Swedish ICUs. EXPOSURES Baseline characteristics, intensive care treatments and organ failures. MAIN OUTCOMES AND MEASURES The primary outcome was 30-day all-cause mortality. A multivariable model was used to determine the independent association between potential predictor variables and death. RESULTS We identified 1563 patients with complete 30-day follow-up. The 30-day all-cause mortality was 26.7%. Median age was 61 [52 to 69], Simplified Acute Physiology Score III (SAPS III) was 53 [46 to 59] and 62.5% had at least one comorbidity. Median PaO2/FiO(2) on admission was 97.5 [75.0 to 140.6] mmHg, 74.7% suffered from moderate-to-severe acute respiratory failure. Age, male sex [adjusted odds ratio (aOR) 1.5 (1.1 to 2.2)], SAPS III score [aOR 1.3 (1.2 to 1.4)], severe respiratory failure [aOR 3.0 (2.0 to 4.7)], specific COVID-19 pharmacotherapy [aOR 1.4 (1.0 to 1.9)] and continuous renal replacement therapy [aOR 2.1 (1.5 to 3.0)] were associated with increased mortality. Except for chronic lung disease, the presence of comorbidities was not independently associated with mortality. CONCLUSIONS Thirty-day mortality rate in COVID-19 patients admitted to Swedish ICUs is generally lower than previously reported despite a severe degree of hypoxaemia on admission. Mortality was driven by age, baseline disease severity, the presence and degree of organ failure, rather than pre-existing comorbidities.