It is common practice to begin anticoagulant treatment of deep-vein thrombosis with a 10-day course of intravenous heparin, with warfarin added on day 5 to 10 and continued for several months. We performed a randomized, double-blind trial comparing a shorter course of continuous intravenous heparin (5 days, with warfarin sodium begun on the first day) with the conventional 10-day course of heparin (with warfarin sodium begun on the fifth day) in the initial treatment of 199 patients with acute proximal venous thrombosis documented by venography. The frequency of objectively documented recurrent venous thromboembolism was low and essentially the same in the two groups (7.1 percent in the short-course group vs. 7.0 percent in the long-course group). Because the observed difference between the groups was 0.1 percent in favor of the long-course group, it is unlikely (P less than 0.05) that a true difference in favor of this group would be greater than 7.5 percent; the difference could be as much as 7.3 percent in favor of the short-course group. Major bleeding episodes were infrequent, and the rate was similar in both groups. We conclude that a five-day course of heparin is as effective as a 10-day course in treating deep venous thrombosis. Furthermore, using the shorter course would permit earlier discharge from the hospital and thus offer substantial cost savings.