The pathogenesis and management of frozen shoulder remain controversial. A review of the literature suggests that constrictive capsulitis is a pathological feature common to chronic cases, and this provides the rationale for manipulative treatment. Thirty patients with frozen shoulders were randomly allocated to two treatment groups. One group received manipulation and local steroid injection and their outcome was compared to the other group treated by local steroid injection alone. Review at one month showed little difference between the two treatment groups. At three months the group treated by manipulation and steroid injection showed good improvement in range of movement (40%) compared to those who received injection alone (13%). The manipulated group also showed substantial improvement in day pain scores (80%) compared to those who received injection alone (47%) and their disability resolved in 47% compared to the injection alone group where only 13% had no disability. The numbers of patients in this trial are too small for the results to be significant statistically but the trends shown are consistent with other similar series and it is suggested that this form of management is both logical and effective.