Indications for angiography in the evaluation of penetrating extremity trauma remain controversial. Our experience was reviewed to determine the yield of angiography in penetrating extremity trauma and to correlate clinical findings with angiographic results. During an 81-month period from 1983 through 1989, 284 extremity arteriograms were carried out in 268 patients. The angiographic yield in patients with abnormal clinical findings was 51%. The angiographic yield in patients when proximity of the injury to major vessels was the only indication was 6% (7% with gunshot wounds and 0% with stab wounds). Neurologic deficit alone as an indication for angiography accounted for 55% of the angiograms interpreted as "negative" and none of those interpreted as "positive". We conclude that the use of angiography in patients with gunshot wounds to the extremity with "proximity injuries" to major vessels should continue, its use is not warranted in extremity stab wounds when proximity is the sole indication, and abnormal neurologic findings in the absence of other findings are a poor predictor of vascular injury.