The subjects included inmates who had been employed in the industry for at least 6 continuous months during the years 1981 and 1982, and industry nonparticipants confined during the same period. Of the 1,434 subjects, 399 industry participants and 497 nonparticipants had been released from from custody by July 1986. Inmate records were examined for 80 percent of the sample, and background information was coded on each subject. This included data admitted to prison, commitment crime, sentence information, age, ethnicity, educational background, military service, preprison employment information, marital status, preprison drug and alcohol use, and institutional misconduct record. Findings indicate that 29 percent of industry participants and 34 percent of nonparticipants were recidivists. The measure of recidivism was any felony arrest following release. Although industry participants had a lower recidivism rate than nonparticipants, when differences between the groups were controlled on other characteristics associated with recidivism, the rates of participants and nonparticipants were virtually identical. A proportional hazards regression model was estimated that showed that, net of other variables, the effect of prison industry participation on the probability of postrelease felony arrest was small and insignificant. The findings are discussed in the context of existing correctional outcome research and recent research and recent developments in prison work programs. Tabular data and 18 references. (Author abstract modified)