The results of randomized clinical trials evaluating commonly used methods of deep vein thrombosis (DVT) prophylaxis in moderate- and high-risk general surgery patients were pooled to obtain an unbiased estimate of efficacy and risks. Low-dose heparin (LDH), dextran, heparin-dihydroergotamine (HDHE), intermittent pneumatic compression (IPC), and graded elastic stockings significantly reduced the incidence of DVT; aspirin was ineffective. In contrast to other methods, elastic stockings have not been adequately studied to determine their value in reducing DVT in high-risk patients, such as those with malignancy. Only LDH and dextran were studied in numbers of patients sufficient for demonstrating a clear reduction in pulmonary embolism (PE). In comparison studies, LDH was superior to dextran in preventing DVT, but the two agents were equivalent in protecting against PE. Although HDHE was marginally better than LDH in preventing DVT, it appeared to have no advantage in preventing PE--at least in moderate-risk patients. The incidence of major hemorrhage was not increased with any of the prophylactic agents. However, wound hematomas occurred significantly more frequently with LDH, an effect noted in the pooled data from double-blind and open trials. In comparison trials with LDH, both dextran and HDHE had significantly fewer wound hematomas. LDH administered every 8 hours appeared more effective in reducing DVT than LDH administered every 12 hours; the incidence of wound hematomas was equivalent with both regimens.