Berk and Rauma collected data on over 1,000 ex-offenders released from California State prisons between July 1978 and December 1980 who applied for unemployment benefits under a program created by a 1977 law. Using time to failure as the measure of recidivism and a proportional hazards model, this study replicated the Berk and Rauma logistic regression results. For all but about 100 cases, at least 10 months of followup data were available. Data were also collected from Ventura County on 50 parolees for 120 days immediately following release. Not everyone in the second sample applied for the unemployment benefits, and the group included civilly committed addicts released from the California Rehabilitation Center. Analysis of the statewide data confirmed Berk and Rauma's findings that the unemployment benefits significantly reduced recidivism and discovered that this effect continued after benefits expired. The Ventura study demonstrated that, when they do work, many offenders work on a regular basis. Again, no subsequent increases in criminality occurred when benefits expired. The data also suggested that the benefits program contained some work disincentives, but acted as an insurance policy against unemployment for working ex-offenders. The Ventura analysis shows that researchers can improve the quality and quantity of their results by analyzing longitudinal instead of cross-sectional data. Tables, graphs, and over 50 references are included.