The medical records of 66 patients who underwent evaluation for possible deep venous thrombosis (DVT) by means of gradient-echo (GRE) magnetic resonance (MR) imaging were reviewed. Confirmatory venograms were obtained in 26 patients; in the other 40, the accuracy of GRE MR imaging was assessed by means of clinical follow-up, which ranged from 1 to 23 months (mean, 7.7 months). Findings in GRE MR images were negative in 42 patients (64%) and positive in 24 patients (36%). In patients who underwent confirmatory venography, the sensitivity of GRE MR imaging was 100% and the specificity was 92.9%. No patient with a negative GRE MR study developed DVT or pulmonary emboli during the follow-up period. Although seven patients in this group died, the single autopsy showed no evidence of pulmonary embolism, and no clinical evidence existed to suggest that pulmonary embolism was the cause of death in any of the other six patients. It is concluded that GRE MR imaging is an accurate, noninvasive means of detecting DVT.