Background: Policies and programs that emphasize employment for persons with mental illness are often promoted with the goals of improving economic self-sufficiency and reducing dependence on public welfare programs. At present, there is little empirical evidence about the actual effect of vocational interventions on economic self-sufficiency or on use of public benefits by persons with mental illness. Study Aims: This study provides a preliminary look at how participating in supported employment, a form of vocational rehabilitation emphasizing ongoing support in competitive jobs, affects the amount that participants earn from work and the total amount of income they receive from all sources. Further, we examine the extent to which receiving public benefits affects the amount earned from private employment, taking into consideration other factors that might be associated with benefit status. Methods: Data are from a randomized trial of supported employment interventions. This analysis followed 137 of those study participants with severe mental illness for 18 months after they enrolled in either of two supported employment programs. Income from various sources was estimated based on interviews with study participants upon study entry and at six-month intervals thereafter. Changes in income from work, government and other sources were analyzed using paired Wilcoxon matched-pairs signed-ranks tests and t-tests. Using ordinary least-squares regression, we analyzed the effect of benefit status on changes in earnings, taking into account diagnosis, work history, education, program type, site of program, psychiatric symptoms, global functioning and previous earnings. Results: Estimated total income increased by an average of $134 (US) per month after enrolling in supported employment. More than three-quarters of this increase was from government sources, such as Social Security and educational grants. The increase in government income was largely due to participants applying for and getting cash benefits for the first time. Social Security payments for those receiving benefits before enrollment did not change significantly. A small group of persons (n = 22) who did not