The efficacy of low-dose heparin in preventing fatal postoperative pulmonary embolism has been investigated in a multicentre prospective randomised trial. 4121 patients over the age of forty years undergoing a variety of elective major surgical procedures were included in the trial; 2076 of these were in the control group and 2045 patients received heparin. The two groups were well matched for age, sex, weight, blood-group, and other factors which could predispose to the development of venous thromboembolism. 180 (4·4%) patients died during the postoperative period, 100 in the control and 80 in the heparin group: 72% of deaths in the control and 66% in the heparin group had necropsy examination. 16 patients in the control group and 2 in the heparin group were found at necropsy to have died due to acute massive pulmonary embolism (P<0·005). In addition, emboli found at necropsy in 6 patients in the control group and 3 in the heparin group were considered either contributory to death or an incidental finding since death in these patients was attributed to other causes. Taking all pulmonary emboli together, the findings were again significant (P<0·005). Of 1292 patients in whom the 125I- fibrinogen test was performed to detect deep-vein thrombosis (D.V.T.) 667 were in the control group and 625 in the heparin group. The frequency of isotopic D.V.T. was reduced from 24·6% in the control group to 7·7% in the heparin group (P<0·005). In 30 patients D.V.T. was detected at necropsy; 24 in the control and 6 in the heparin group (P<0·005). 32 patients in the control group and 11 in the heparin group developed clinically diagnosed D.V.T. which was confirmed by venography (P<0·005). In addition, 24 patients in the control and 8 in the heparin group were treated for clinically suspected pulmonary embolism. The difference in the number of patients requiring treatment for D.V.T. and/or pulmonary embolism in the two groups was again significant (P<0·005). 9 patients were found at necropsy to have died from hæmorrhage; 5 were in the control and 4 in the heparin group. A careful objective analysis of operative and postoperative bleeding in 1475 patients showed no statistically significant difference in the blood-transfusion requirements or in the fall in the postoperative hæmoglobin level either in the individual operative groups or in the group as a whole. However, the difference in the number of patients who developed wound hæmatoma in the heparin and control groups was significant (P<0·01). The results of the trial indicate that this form of prophylaxis can now be recommended for use on a large scale in " high-risk " patients undergoing major surgery.