Both the experimental group (dischargees) and the control group (parolees) were released from Connecticut prisons in 1973-74. Each group was followed for periods of 1 year, 2 years, and 3 years to determine whether parole made a difference in recidivism rates. After 1 year, 39 percent of the 57 parolees had failed (new conviction), while 67 percent of the 111 dischargees had failed. Parolees also performed somewhat better than dischargees on the variable 'time in the community' (the percentage of the followup year the subject spent in the community as compared with time spent in jail, prison, or mental hospital). After 3 years, 77 percent of the parolees had failed, while 85 percent of the dischargees had failed. This eight-point difference was not significant. Thus, an 8-month period on parole had no crime-preventing effects after either 2 or 3 years following release. Multivariate analysis showed that none of the correlations between type of release and recidivism was significantly affected when tested for the effect of 20 variables on outcome. Policy implications of the findings are discussed. A discussion of the multivariate analysis is appended, and tabular data and footnotes are provided.