Intraoperative venodilation in veins distant from the site of operation has been shown to occur in animals and has been directly correlated with focal venous endothelial damage. This exposure of subendothelial collagen could serve as initiation sites for thrombus formation. This study tests the hypothesis that human beings (1) significant operative venodilation occurs and that it correlates with postoperative deep venous thrombosis (DVT); (2) operative venodilation can be pharmacologically controlled; and (3) this control reduces the incidence of postoperative DVT. Twenty-one patients undergoing total hip replacement had their contralateral cephalic vein continuously monitored with modified ultrasonographic instrumentation, with a continuous on-line recorder graphing venous diameter. Patients were randomly assigned to receive 0.5 mg of dihydroergotamine and 5000 U of heparin (DHE/Hep) for prophylaxis or placebo, with investigators "blinded" Postoperatively, all patients underwent ascending phlebography. Patients in whom postoperative DVT developed (11) had a mean operative venodilation of 28.9% +/- 3.93%, and those in whom DVT did not develop (10) had a mean venodilation of 11.6% +/- 1.55% (p = 0.001). Only 17% (2/12) dilating less than 20% baseline diameter had DVT compared with 100% (9/9) dilating greater than 20% of baseline diameter (p = 0.002). Patients receiving venotonic agent DHE had significantly less venodilation and DVT (p less than 0.001) compared with patients receiving the placebo. Patients who had DVT and whose veins dilated greater than 20% were older than patients who did not have DVT and whose veins minimally dilated: p = 0.04 and p = 0.07, respectively. Although there was a trend toward increased venoconstriction in patients receiving DHE/Hep (p = 0.09), there was no correlation of venoconstriction with ultimate thrombotic outcome. Maximal venodilation occurs during handling of soft tissue (muscle), and this occurs significantly sooner than maximal venoconstriction, which occurs during bone manipulation. We conclude that excessive operative venodilation is a new and important etiologic factor that leads to postoperative DVT. Operative venodilation can be pharmacologically controlled with the venotonic agent DHE. The combination DHE/Hep reduces postoperative DVT by the reduction of operative venodilation in the presence of low doses of an anticoagulant. These findings offer a new approach for predicting postoperative DVT and an object rationale for developing effective prophylaxis.