Four treatments for sciatic symptoms--traction, exercises, manipulation, and corset--were assessed in a randomised controlled trial in 322 outpatients. The design was factorial. There were thus sixteen treatment groups, enabling a comparison of combinations of methods as well as of individual methods. Treatment lasted for four weeks. Patients were reviewed at the end of this period and at four and sixteen months after entry to the trial. Progress was measured by the patient's account of symptomatic improvement or deterioration and by return to work or normal activities. At four weeks each of the treatments was associated with a small degree of benefit over and above the high rate of spontaneous improvement. For manipulation, the benefit was statistically significant on one of the scales used to measure progress. There was a significant increase in symptomatic improvement with increasing numbers of treatments used in combination. This was complemented by a clear tendency for those who had received fewer types of treatment during the trial to have further treatment in the ensuing three months. There were no beneficial effects of treatment detectable at four or sixteen months. In the short-term, active physiotherapy with several treatments appears to be of value in the outpatient management of patients with sciatic symptoms, but it does not seem to confer any longer-term benefit.