Abstract: Purpose: To explore the intensity and variation of workers’ worries, pain, psychosocial factors, and margin of manoeuvre before and after a return-to-work program, and identified the psychosocial factors associated with non-return to work at the end of the rehabilitation program. Methods: A pre-post study design was used. A convenience sample of 80 workers starting a return-to-work program and having a compensated musculoskeletal injury that caused an absence of more than three months from their regular work was recruited. Data were collected at baseline and at the end of the rehabilitation program on the nature of the worries and maintenance factors defined in Dugas’ generalized anxiety and worry model, using validated questionnaires. The margin of manoeuvre was assessed by the treating occupational therapist. A series of descriptive analyses were performed, as well as Generalized Estimating Equations analyses. Results: Workers’ worries were work-related or disability-related 83% of the time at baseline. These worries were essentially based on the situation then occurring at work 90% of the time. For the Generalized Estimating Equations analyses on work status, the final model was significant, explaining 54% of the variance in non-return to work (Pseudo R2 = 0.54; p = 0.0001). Workers were 8.52 times less likely to return to work when the margin of manoeuvre was insufficient, and twice as likely not to return to work in the presence of intense worry. Worries were significantly associated with insufficient margin of manoeuvre. Conclusion: A strong association between workers’ lack of margin of manoeuvre at work and their worries about their return to work, and poor work outcomes, supports the importance of the worker-environment interaction in rehabilitation programs.