The purpose of this study was to assess the efficacy of back school programmes for low-back pain. Data sources comprised computer-aided search of published randomized clinical trials and assessment of the methods of the studies. 21 papers reporting on 16 randomized clinical trials evaluating a back school programme were selected for the study. Data extraction included a score for quality of the methods, based on four categories: study population, interventions, effect measurement and data presentation and analysis; and the conclusion of the author(s) with regard to the efficacy of the back school programme. Only two studies scored more than 50 points (maximum = 100 points) indicating the overall poor quality of the methods. Seven studies indicated that the back school programme was more effective than the reference treatment and seven reported it to be no better or worse than the reference treatment. In two studies the authors refrained from drawing a conclusion. The studies reporting positive results showed higher methods scores (4/7 positive vs 0/7 negative scored > or = 45 points). Reported benefits of back schools were usually of short duration only. There are major flaws in the design of most studies. The best studies indicated that back schools may be effective in occupational settings in acute, recurrent or chronic conditions. The most promising type of interventions were (modifications of) the "Swedish back school" and were quite intensive (a 3 to 5-week stay in a specialized centre). Future research efforts should focus on the identification of patients who would benefit most from back schools. In addition, more attention should be paid to the cost-effectiveness of back schools.