All cases of penetrating extremity trauma (PET) seen at an urban trauma center were prospectively studied to determine the accuracy and safety of physical examination as the sole mode of evaluation for vascular injury. All patients with PET producing obvious or "hard" signs of vascular injury underwent immediate surgery. All asymptomatic proximity wounds were observed in hospital for 24 hours before discharge to outpatient followup. Patients with non-proximity wounds were discharged immediately. Patients with shotgun wounds and thoracic outlet injuries also underwent arteriography. Of 2,674 trauma patients evaluated during the 1-year study period, 310 (11.6%) had 366 penetrating extremity wounds, most (71%) occurring in the lower extremities. Gunshots caused most (82%) of the wounds, followed by stabs or lacerations (14.5%), and shotguns (3%). Clinically occult wounds in proximity to major limb vessels were the most common (78%) category of clinical presentation, followed by asymptomatic non-proximity wounds (16%), and wounds producing hard signs (6%). There were two missed vascular injuries, both in the asymptomatic proximity group (0.7% false negatives). Every patient taken immediately to surgery for hard signs had major arterial injury requiring repair, for a 100% positive predictive value for physical examination. No mortality or morbidity were related to protocol management. These results to date support prior reports of a negligible incidence of significant vascular injury following clinically occult proximity PET, and further suggest that the overall predictive value of physical examination of PET for vascular injury approaches 100%.