We sought to prospectively validate an algorithm for selective radiography in the Emergency Department (ED) management of patients with suspected shoulder dislocation. Physicians ordered pre- and post-reduction radiographs based on an algorithm incorporating the mechanism of injury, previous dislocations, and the physician's clinical certainty of joint position. Follow-up consisted of telephone calls and chart review. Of 100 patients, 94 had shoulder dislocation, and 59% were recurrent. Thirty percent had both pre- and post-reduction radiographs, 45% had pre- or post-reduction only, and 25% had no shoulder films, yielding an overall 46% reduction in X-ray utilization. Mean ED times were significantly shorter for patients managed without radiographs. On telephone follow-up (76%) and chart review (100%), we found no missed fractures or persistent dislocations. Use of a clinical decision rule for selective radiography reduced the number of radiographs and time spent in the ED, while missing no fractures or dislocations.