This study prospectively evaluated the outcome of manipulation under anaesthesia and hydrodilatation as treatments for adhesive capsulitis. A total of 36 patients (38 shoulders) were randomised to receive either method, with all patients being treated in stage II of the disease process. The mean age of the patients was 55.2 years (44 to 70) and the mean duration of symptoms was 33.7 weeks (12 to 76). Eighteen shoulders (17 patients) underwent manipulation under anaesthesia and 20 (19 patients) had hydrodilatation. There were three insulin-dependent diabetics in each group. The mean visual analogue score in the manipulation under anaesthesia group was 5.7 (3 to 8.5; n = 18) before treatment, 4.7 (0 to 8.5; n = 16) at two months (paired t-test p = 0.02), and 2.7 (0 to 9; n = 16) at six months (paired t-test, p = 0.0006). The mean score in the hydrodilatation group was 6.1 (4 to 10; n = 20) before treatment, 2.4 (0 to 8; n = 18) at two months (paired t-test, p = 0.001), and 1.7 (0 to 7; n = 18) at six months (paired t-test, p = 0.0006). The visual analogue scores in the hydrodilatation group were significantly better than in the manipulation under anaesthesia group over the six-month follow-up period (p < 0.0001). The mean Constant score in those manipulated was 36 (26 to 66) before treatment, 58.5 (24 to 90) at two months (paired t-test, p = 0.001) and 59.5 (23 to 85) at six months (paired t-test, p = 0.0006). In the hydrodilatation group it was 28.8 (18 to 55) before treatment, 57.4 (17 to 80) at two months (paired t-test, p = 0.0004) and 65.9 (28 to 92) at six months (paired t-test, p = 0.0005). The Constant scores in the hydrodilatation group were significantly better than in the manipulated group over the six-month period of follow-up (p = 0.02). The range of movement improved in all patients over the six months, but was not significantly different between the groups. At the final follow-up, 94% of patients (17 of 18) were satisfied or very satisfied after hydrodilatation compared with 81% (13 of 16) of those receiving a manipulation. Most of our patients were treated successfully, but those undergoing hydrodilatation did better than those who were manipulated