Purpose. This study investigates factors that will affect the chances of long-term sick workers returning to work.

Method. The study is based on a representative sample of persons with mental illnesses and musculoskeletal disorders who received a rehabilitation allowance in Norway in 2002. Their occupational status in 2005 and experiences from the rehabilitation process were charted through their responses to a questionnaire in combination with national register data. Our analytical framework is based on three domains: the medical domain, the domain of authoritative bodies and the production domain (working life), as described by Lindqvist (Att satta granser: organisationer och reformer i arbetsrehabiliteringen. Umea: Borea; 2000).

Results. Our main findings are that earlier work experience, age and other members in household, in addition, to influence over one's own rehabilitation process explains whether workers on long-term sick leave return to work. Furthermore, individual factors such as gender and diagnosis (i.e. musculoskeletal disorder and mental illness) do not explain the probability to return to work. Experience and contact with representatives of the medical, the authoritative bodies and the production domain only explain to a small extent the probability to return to work. It is even interesting to note that regular contact with the social insurance office (SIO) has a negative effect on the probability to return to work. This may indicate that contact with the SIO subjects sick workers to a risk of developing a 'social insurance career' to secure their income.

Conclusions. The findings show that the medical, the authoritative bodies and the production domain might represent different logics that can pull a sick worker in different directions during the rehabilitation process. However, these different logics do not fully explain, which outcome a rehabilitation process takes. It is important to extend the analysis to include how the individuals respond to these logics during the rehabilitation process.