The City of Winnipeg's community services programs (CSPs) are examples of the "work experience" approach to helping social assistance recipients find employment. Like other evaluations of work experience programs, this assessment finds the CSPs effective and cost-beneficial, in the short run, in reducing dependency on social assistance. Unlike most program assessments, however, this evaluation measures participants' length of involvement in the program and finds there is an optimal duration of involvement. A quasi -experimental design is used to assess the program's net impact on participants' subsequent length of stay on assistance. Self-selection bias is assessed, and multiple-regression analysis is used to control for differences between program and comparison group members and to determine the unique impact of program participation and length of participation.