A group of 111 surgical patients at high risk of venous thrombosis were studied after operation by independent clinical assessment and with 125 I-fibrinogen to detect venous thrombosis. Almost half of the patients developed venous thrombosis. Of these, two-thirds were not suspected clinically despite careful scrutiny. In the patients in whom a clinical diagnosis of venous thrombosis was made this diagnosis was falsely positive in a quarter. More than half of all thrombotic episodes were detectable on the day after operation. The prevalence of venous thrombosis, together with the difficulty in diagnosing it, strongly supports the argument that a reduction in the incidence of pulmonary embolism must depend on widespread adoption of effective prophylaxis, especially in the large number of patients at high risk of venous thrombosis. Prophylactic trials must be objectively assessed, and it is in this field that the 125 I-fibrinogen technique probably has the most to offer.