In general, randomized controlled studies concerning return to work have failed to demonstrate significant treatment effects for long-lasting musculoskeletal pain, and most treatments examined have not been economically beneficial. Individuals (n=654) sick-listed for at least 8 weeks with musculoskeletal pain, selected from the Norwegian mandatory sickness insurance system and volunteering to participate, were categorized into three groups differing in a prognosis score (good, medium, poor) for return to work, based on a brief, standardized screening of psychological and physiotherapy findings. They were then randomly assigned to three outpatient treatments with three different levels of intensity (ordinary treatment, light multidisciplinary, and extensive multidisciplinary treatment). The evaluation was based on 14 months follow-up data on return to work collected from social security records. The patients with good prognosis for return to work do equally well with ordinary treatment as with the two more intensive treatments. The patients with medium prognosis benefit equally from the two multidisciplinary treatments. The patients with poor prognosis receiving extensive multidisciplinary treatment returned to work at a higher rate than patients with poor prognosis receiving ordinary treatment, 55 vs. 37% (P<0.05) at 14 months. Multidisciplinary treatment is effective concerning return to work, when given to patients who are most likely to benefit from that treatment. Measures of pain or quality of life are not included in this study. The cost-benefit analysis of the economic returns of the light multidisciplinary and the extensive multidisciplinary treatment programs yields a positive net present social value of the treatment. A simple, standardized, screening instrument including only psychological and physiotherapeutic observations may be a useful clinical tool for allocating patients with musculoskeletal pain to the right level of treatment.