The treatment of patients with penetrating extremity trauma in proximity to major arteries as well as the nonoperative treatment of clinically occult arterial injuries remain controversial. Duplex ultrasonography (DUS) has recently been advocated in this setting. We therefore studied experimentally induced arterial injuries in dogs to correlate the natural history, duplex findings, and histopathologic condition of different injuries and to help define criteria for operation. Fifty-two canine femoral and carotid arteries were randomized to have surgically created intimal flaps (n = 15), crush injuries (n = 15), or lacerations (n = 15) or to be controls (n = 7). An experienced sonographer, blinded to the presence or type of injury, evaluated the vessels every 10 days for 1 month. Histopathologic study was performed 1 month after injury when the arteries were retrieved. The sensitivity (96.5%), specificity (86.4%), and accuracy (95.1%) of DUS in evaluating arterial injuries at 1 month correlated well with histopathologic evaluation. All arteries subjected to crush injuries showed abnormal duplex findings. These findings correlated well with the histologic picture of severe injury (arterial wall thickness = 2.72 x +/- 0.23 x control; intramural hemorrhage, 93%; mural thrombus, 60%). DUS and histologic study revealed healing of intimal flaps in 27% of the arteries. Other intimal flaps deteriorated (stenoses, 47%; dilation, 13%; occlusion, 13%). Most lacerations (86%) revealed duplex evidence of healing within 10 days of injury. This was confirmed by histologic study at 1 month in 73% of lacerated arteries. This study confirms the accuracy of DUS in diagnosing various arterial injuries and shows that the natural history of these injuries varies with the mechanism of injury.(ABSTRACT TRUNCATED AT 250 WORDS)