Although indirect support can be found for the clinical benefits of work, it has not been studied in randomized designs, nor have critical variables been manipulated. One such variable is pay incentive. The authors present a study of 150 subjects with schizophrenia or schizoaffective disorder who were randomized into Pay ($3.40/hour) and No-Pay conditions and offered 6-month work placements within a Department of Veterans Affairs medical center. Subjects participated in a work-related support group and were evaluated weekly on symptom measures. Results indicated that Pay subjects worked more hours and earned more money than No-Pay subjects. Pay subjects showed more total symptom improvement at followup, and more improvement, particularly on positive and emotional discomfort symptoms. They also had a significantly lower rate of rehospitalization than No-Pay subjects. Participation in work activity was closely associated with symptom improvement. Participators showed more total symptom improvement at followup than partial participators or nonparticipators, and more improvement, particularly on positive, hostility, and emotional discomfort symptoms. We concluded that pay increased participation and that, in this study, participation in work activity was primarily responsible for symptom reduction.