The goals of the current study were to determine those preprogram (=prognostic) variables and treatment-related changes that predict return to work in the multimodal management of chronic back pain. The outcome measures for 143 patients at 6-month follow-up were analyzed. The program had a duration of 4 weeks, was based largely on the functional restoration approach (Mayer and Gatchel, 1998), and occurred within a workers' compensation framework. Some 87% of the patients successfully returned to work. Three sets of predictor variables were considered: demographic/socioeconomic data, physical measures, and psychological measures. Three prognostic variables proved to be significant negative predictors of return to work: time off work, previous spinal surgery, and a clinically elevated (preprogram) score on the MMPI-2 scale Lassitude-Malaise (Hy3). A repeated-measures MANOVA showed an incomplete return to work to be associated with only limited improvement in self-reported disability and pain report. However, patients who failed to return to work did not differ with regard to improvement in objective physical functioning or psychological distress. It is therefore hypothesized that a change in the perceived disability status is the key element necessary to return patients with chronic back pain to work, although ongoing reinforcement schemes operative in the home work environment may lead to a relapse in the post treatment phase.