We studied prospectively 100 patients (110 shoulders) with primary (idiopathic) frozen shoulder, of whom 22 (22%) were diabetics (30 shoulders). Eighty-eight patients (98 shoulders), whose shoulders had failed to improve after conservative treatment, were then managed either by manipulation under anaesthesia (MUA) alone, or MUA with one of two types of intra-articular injection (methylprednisolone, or a large volume [50100 cc] of normal saline), followed by physiotherapy. Patients were followed for an average of 6-8 months. Our findings showed that patients who had MUA with an intra-articular normal saline injection had better results than those who had MUA either alone or with an injection of steroid. We also noticed a high incidence of failure among diabetic patients with frozen shoulder. The manipulative procedure combined with injection of normal saline was safe and effective, and all materials required are readily available and inexpensive.