The present study investigated work-related determinants of return to work. Our hypothesis was based on the strain hypothesis of the Demand-Control-Support model, which postulates a relation between job demands, job control and support at work on the one hand, and the aetiology of health complaints on the other hand. High demands were hypothesized to obstruct return to work, whereas high control and high support were thought to have a positive effect on return to work. This hypothesis was tested in a population of employees who were sick-listed for 6-8 weeks. Return to work, as operationalized by the categories (i) not working; (ii) return to work with adjustments; and (iii) full return to work, was determined 4 months after the onset of the sick leave. The hypothesis was tested by logistic regression analyses. High job demands were the least predictive of full return to work. However, the likelihood of employees with high job demands returning to work with adjustments was higher than the likelihood of them not working. Therefore, job demands might also work as a pressure to return to work (compare this with Smulders and Nijhuis, 1999). Furthermore, high skill discretion in combination with high job demands predicted working with adjustments in comparison with not working. Finally, high supervisor support was the most predictive of return to work without adjustments, and the least predictive of not working.