A randomized double-blind trial of prophylaxis of thromboembolism in surgical patients judged by clinical morbidity has been completed. Altogether 831 patients over 40 years of age who underwent elective surgery of the stomach, biliary system, or colon received either dextran 70 or normal saline before the operation. Thirteen of the 435 patients on saline and three of the 396 on dextran developed pulmonary embolism. Eight of these 16 patients died of pulmonary embolism--seven in the saline group and one in the dextran group. As detected either clinically or by 125I-fibrinogen scanning the incidence of deep vein thrombosis was similar in the two groups. There was no increased incidence of excessive bleeding in patients on dextran though clinical impression suggested that some patients on dextran bled excessively. This trial showed that dextran 70 administered by intravenous drip during operation is effective in preventing pulmonary embolism and, in particular, reducing mortality from this cause. It seems to be as effective as subcutaneous heparin but is easier to administer and places less of a burden on nursing services.