Corticosteroids are commonly used in the treatment of the impingement syndrome. Efficacy, as well as accurate placement, have been questioned. The purpose of this prospective, randomized study is to assess the accuracy of subacromial injections and to correlate accuracy with short term clinical outcome at 3 months. Sixty shoulders, which were diagnosed with impingement syndrome, were randomized to receive a subacromial injection of corticosteroids, local anesthetic, and contrast dye from 1 of 3 locations: anterolateral, lateral, or posterior. Accuracy was confirmed by 3 radiographic views of the shoulder, while clinical ratings were assessed by the UCLA shoulder score and a 10-point visual pain analog scale during the initial, post-injection, and 3-month visits. The overall accuracy was 70%, with no difference among the 3 portals. Accuracy was not related to body mass index. Furthermore, accurate injections did not significantly improve the UCLA score, pain scale, or patient satisfaction at 3 months. In contrast, accurate injections produced a positive Neer's impingement test more often (35/39 vs 9/16; P = .009). Overall, there was an improvement in the UCLA score (26.2-32.2; P < .001) and a decrease in the pain scale (7.2-3.43; P < .001) at 3-month follow-up. In conclusion, the accuracy of injection was 70%. Clinical improvement did not correlate with accuracy; however, accuracy did reliably produce a positive impingement test. This multimodal treatment plan did produce significant improvement in shoulder function and pain level in the short term.