Introduction: Patients who develop severe COVID-19 have a high risk of thromboembolic events, including both venous and arterial thromboembolism. To mitigate those complications, a broader use of anticoagulant agents is proposed, but this strategy may be associated with bleeding. Methods: We retrospectively collected data of consecutive 246 adult patients admitted to the ICU of Hospital BP - A Beneficência Portuguesa de São Paulo with a confirmed diagnosis of COVID-19 between March 17th and June 18th, 2020. Laboratorial and clinical data regarding thrombosis and hemostasis were evaluated. We then analyzed the incidence of bleeding and venous and arterial thromboembolic events and its correlation with death and mechanical ventilation (MV). Results: The median age was 63 years (range 20-102), and 58.9% were male. At ICU admission, median Simplified Acute Physiology Score 3 (SAPS-3) was 50 (24-100) and 17.4% of patients had a d-dimer at admission higher than 3,000 ng/mL. During ICU stay 113 patients (45.9%) required MV and 231 patients (94.3%) received anticoagulants (18.3% therapeutic doses). Fifty-one (20.8%) patients had thrombocytopenia defined as platelets lower than 100,000/uL during the study period. Median fibrinogen level was 437 mg/dL. Eighty-three patients died during the 28 days of follow-up (33.7%), and the cause of the death was attributable to thromboembolic events in 5/83 (6.0%). The incidence of any thromboembolic event was 16.7%. Venous thromboembolism (VTE) occurred in 11.6% of the patients and arterial thrombosis in 6.5% (stroke in 2.1%, myocardial infarction in 4.9%, limb ischemia in 0.8%). Major bleeding was observed in 7 patients (2.9%) and 5 of them died. Mortality of patients who had a VTE event was 42.8%. MV requirement, thrombocytopenia and higher d-dimer levels were correlated with death on univariate analysis. Discussion and conclusion: Our study showed a lower incidence of VTE than previously reported in European ICU cohorts of COVID-19 patients and this may be attributable to a high adhesion to pharmacological venous thromboprophylaxis, although underdiagnosis may have influenced our results (venous ultrasound was performed in 6.9% of patients, CT pulmonary angiography in 5.3% and echocardiogram in 28.3%). The bleeding frequency of 2.8% was like the one reported by a multicentric US study. Development of institutional protocols of VTE prophylaxis and a lower threshold to perform diagnosis tests for VTE in COVID-19 patients could improve this results. The optimal strategy to prevent thromboembolic events in this context is still a matter of debate.