Approximately 25 percent of the sample was placed in halfway houses either voluntarily or involuntarily. The recidivism patterns of those placed in halfway houses were compared with those not residing in halfway houses. Those placed in halfway houses tended to have drug-alcohol abuse histories and extensive criminal histories, suggesting that their placement was intended to provide treatment and close supervision. Even though halfway house residents were generally poorer risks than nonresident releasees, their recidivism rate was a percentage point less than nonresidents. Halfway house residents, however, had significantly higher absconder rates and trouble rates (arrests, violation reports, and warrants that did not result in return to prison). Halfway houses were particularly effective in curtailing the recidivism rates of alcohol abusers, the chronically unemployed, and assaultive offenders. Halfway houses had only moderate success in controlling recidivism for property offenders, those with no history of alcohol/drug abuse, and those considered to be fair to good risks. Halfway houses were notably ineffective in reducing the recidivism of those with drug abuse histories and stable employment. Tabular and graphic data.