90 patients with primary frozen shoulder were randomized to receive an intra-articular glenohumeral corticosteroid injection via a rotator interval approach (n=45) or a posterior approach (n=45). The outcomes of interest included pain on a Visual Analog Scale (VAS), the American Shoulder and Elbow Surgeons (ASES) score, the subjective shoulder value (SSV), and range of motion (ROM). Outcomes were assessed at 3, 6 and 12 weeks post-injection. Improvements in all clinical outcomes were observed between the two groups. No differences were observed between the rotator interval and posterior approach groups, with the exception of the improvement in range of motion in forward flexion and abduction, which was superior in the rotator interval group. The rate of injection success was, however, lower in the rotator interval group compared to the posterior approach group.