Three treatments for non-specific lumbar pain--balneotherapy, underwater traction bath, and underwater massage--were assessed in a randomised prospective controlled trial in 158 outpatients. Each group was treated for four weeks and patients were reviewed at the end of this period and at 12 months after entry to the trial. The prescription of analgesics and the pain score were significantly reduced in all three treated groups, but there was no difference between the three groups. No significant change occurred in spinal motion and the straight leg raising test. After one year only the analgesic consumption was significantly lower than in the control group.