We performed a prospective study of 317 patients in order to determine the prevalence of deep venous thrombosis after reconstructive operations on the spine; 126 of the patients were examined with duplex ultrasound assessments of the lower extremities to ensure that no asymptomatic thrombi were being missed. Thigh-high stockings and sequential pneumatic compression of the lower extremities were used, in all patients, for prophylaxis against venous thrombosis. No antiplatelet agents or anticoagulant medications were administered. There was no evidence of thrombosis on any of the duplex ultrasound studies. Subsequently, venous thrombosis developed and was treated successfully in one of the 126 tested patients and in one of the 191 untested patients, and a fatal pulmonary embolus developed in one of the untested patients. The over-all clinical prevalence of thrombotic complications was 0.9 per cent (three complications in 317 patients). All three of the patients who had clinical evidence of thrombosis had had an anterior lumbar procedure because of a degenerative disorder or trauma; however, we could not prove that this approach or these diagnoses were significant risk factors for thrombosis (p < 0.05). While it is possible that some thrombi may have escaped both clinical and ultrasonic detection, such thrombi apparently were not enough of a danger to warrant the use of intensive prophylactic procedures that are associated with more risk. On the basis of this prospective study, therefore, we think that routine screening for the detection of asymptomatic thrombosis in patients who have had a procedure on the spine is unwarranted.