A common mode of deep vein thrombosis prophylaxis in medical inpatients is unfractionated heparin 5000 U subcutaneously (s.q.) twice daily. We examined the evidence in favor of using this dose of heparin in this group of patients. MEDLINE was searched for studies using the words deep vein thrombosis prophylaxis and heparin. All randomized controlled trials comparing heparin and placebo or heparin and a low molecular weight heparin were used. Relative risk was 0.4 (95% confidence interval 0.22-0.73) in studies comparing heparin 5000 U s.q. b.i.d. with placebo. Relative risk was 0.28 (95% confidence interval 0.21-0.38) in studies comparing heparin 5000 units s.q. t.i.d. versus placebo. In studies comparing unfractionated heparin with enoxaparin relative risk was 1.42 (95% confidence interval 0.99-2.05). Heparin 5000 U s.q. b.i.d. is less efficacious than low molecular weight heparins and unfractionated heparin 5000 U s.q. t.i.d.