In a single-blind trial, five treatments for painful stiff shoulder were compared for a 4-week assessment period in 60 patients. The treatments were acupuncture, steroid injection with placebo and with active tolmetin sodium, physiotherapy in the form of ultrasound and 'placebo' physiotherapy with placebo tolmetin sodium. Objective assessment was gained by use of goniometer readings to monitor shoulder abduction. Pain was measured by visual analogue scales and by a 4-point scale. Comparative assessment was also recorded and at the end of the study a success or failure was recorded for each patient's treatment. With very few exceptions all patients improved markedly, both in terms of the subjective and objective parameters. No differences between the treatments were detected. The incidence and severity of side-effects was low. It is suggested that the results show that the painful stiff shoulder may be a self-limiting condition and that any beneficial effect was really due to natural recovery. This is an important consideration because patients do not always receive immediate attention when referred to an out-patient department and the use of physiotherapy and acupuncture in such cases, perhaps, should be critically examined.