A randomized controlled trial was used to evaluate an educational booklet on back pain for patients presenting to five group practices during one calendar year. The booklet had no immediate effect on consultations for back pain, but in the period from two weeks to one year after presentation significantly fewer patients in the group receiving the booklet consulted with back pain (35.6%) than in the control group (42.2%) (P less than 0.05). There were no significant differences between the booklet and control groups in certified absence from work owing to back pain. Referral to hospital, referral to physiotherapy, admissions to hospital and laminectomies were all less common in the booklet group. The reduction in the combined referral rate to physiotherapy and hospital, and the reduction in laminectomy rate almost reached statistical significance at the 5% level. In replying to a questionnaire sent one year after entry to the study 94.1% of respondents in the booklet group said that they had read the book, 84.0% said that they found it useful, and 68.0% said that they still had a copy. Scores on a 15-item test of knowledge about back pain were significantly higher in the group of patients who had received the booklet than in the control group. The results suggest that the booklet had some effect in altering both the knowledge and behaviour of patients with back pain. The provision of an educational booklet was a method of giving information which was appreciated by both patients and doctors.