The long-term outcome results of inpatient and outpatient treatment of low back pain (LBP) were studied in 476 subjects (aged 35-54, 63% men) randomly assigned to three study groups: inpatients (n = 157), outpatients (n = 159), and controls (n = 160). The study included changes in the severity of low back pain, grade and disability, compliance with self-care, data on disability pensions, and days of sickness allowance during a 2.5-year follow-up period. These variables were used as outcome criteria. Pain and disability had decreased significantly in the two treated groups up to the 3-month follow-up. LBP was still a little slighter in the inpatients at the 1.5-year and 22-month follow-ups, but there were no significant differences between the groups in disability caused by LBP. The refresher programme carried out 1.5 years after the first one did not bring about as clear short-term improvement in pain and disability as the first treatment. During the whole 2.5-year follow-up compliance with self-care was better in the two treated groups, especially in the inpatients. Days of sickness allowance had increased somewhat more in the controls than in the inpatients during the follow-up. No differences between the groups were found in the number of disability pensions granted.