Background A relatively high mortality of severe coronavirus disease 2019 (COVID-19) is worrying, and the application of heparin in COVID-19 has been recommended by some expert consensus because of the risk of disseminated intravascular coagulation and venous thromboembolism. However, its efficacy remains to be validated. Methods Coagulation results, medications, and outcomes of consecutive patients being classified as having severe COVID-19 in Tongji hospital were retrospectively analyzed. The 28-day mortality between heparin users and nonusers were compared, as was a different risk of coagulopathy, which was stratified by the sepsis-induced coagulopathy (SIC) score or D-dimer result. Results There were 449 patients with severe COVID-19 enrolled into the study, 99 of them received heparin (mainly with low molecular weight heparin) for 7 days or longer. D-dimer, prothrombin time, and age were positively, and platelet count was negatively, correlated with 28-day mortality in multivariate analysis. No difference in 28-day mortality was found between heparin users and nonusers (30.3% vs 29.7%, P = .910). But the 28-day mortality of heparin users was lower than nonusers in patients with SIC score >= 4 (40.0% vs 64.2%, P = .029), or D-dimer >6-fold of upper limit of normal (32.8% vs 52.4%, P = .017). Conclusions Anticoagulant therapy mainly with low molecular weight heparin appears to be associated with better prognosis in severe COVID-19 patients meeting SIC criteria or with markedly elevated D-dimer.