OBJECTIVE: Experts do not agree on what, if any, relationships exist between diagnosis, symptomatology, work skills, and the future vocational performance of persons with severe mental illness. The objective of this study was to longitudinally examine such relationships, using a sample of clients who were attending psychosocial rehabilitation programs. METHODS: Subjects were 275 clients of three psychosocial rehabilitation programs who had expressed a vocational goal. They were assessed at intake into the study and then quarterly until they left the rehabilitation program. The variables examined included symptoms, measured by the Brief Psychiatric Rating Scale; diagnosis; work skills, measured by the Griffiths Work Behavior Scale; and vocational status at end-point. RESULTS: Among subjects remaining in the study for one year, both symptomatology and work skills improved significantly. Moderately significant negative correlations were found between symptoms and work skills; subjects who became employed had lower symptom scores and higher work skills than persons who never became employed. CONCLUSIONS: Although a moderate relationship was found between symptomatology and work skills, symptoms should not be considered a proxy measure for vocational functioning among persons with severe mental illness. Participation in psychosocial rehabilitation programs appeared to have a salutary effect on symptoms and work skills.