HYPOTHESIS Frozen shoulder has a greater incidence, more severe course, and resistance to treatment in patients with diabetes mellitus compared with the general population. We hypothesized that diabetic patients with frozen shoulder undergoing treatment with manipulation under general anaesthesia (MUA) would have the same outcome as patients without diabetes. MATERIALS AND METHODS We retrospectively analyzed data collected during a 10-year period of referrals for frozen shoulder. In all cases, a standardized MUA protocol was followed once the diagnosis of frozen shoulder in the frozen phase was made; this included an early repeat MUA in individuals with recurrence. We compared outcomes for patients documented as having diabetes with a nondiabetic control group and assessed the effect of insulin dependence and frozen shoulder etiology within the diabetic group. RESULTS Of a consecutive series of 315 frozen shoulders, 36 patients (39 shoulders) were included in the diabetic group, with 256 patients (274 shoulders) as controls. There was a significant improvement in range of movement and Oxford Shoulder Score (P all <.001), with no difference between diabetic and control groups at early or late follow-up (mean, 41 months). A repeat procedure was required in 36% of diabetic patients compared with 15% of control patients. Recurrence in the diabetic group was influenced by etiology (47% of primary vs 0% of secondary frozen shoulders) and insulin requirement (39% insulin-dependent vs. 31% non-insulin-dependent). CONCLUSION We provide a strategy for the management of diabetic frozen shoulders using MUA and estimates of success and recurrence rates that may be useful when informing consent.