STUDY SUBJECTS WERE 205 MEN AND WOMEN WHO RESIDED IN THE FACILITIES FOR AT LEAST 21 DAYS FROM JANUARY 1973 TO JANUARY 1975. ALL OF THE SUBJECTS HAD EXTENSIVE CRIMINAL RECORDS, NO JOB SKILLS OR EMPLOYMENT POSSIBILITIES, AND INADEQUATE LIVING ARRANGEMENTS. THEY WERE REGARDED BY THEIR PROBATION AND PAROLE OFFICERS AS HIGH RISK. THE SUBJECTS INCLUDED PROBATIONERS (45 PERCENT), PAROLEES (20 PERCENT), INMATES ON PRERELEASE (17 PERCENT), AND PERSONS UNDER OBSERVATION (18 PERCENT). DURING THEIR STAYS AT THE HALFWAY HOUSES, THEY HAD PARTICIPATED IN A TOKEN ECONOMY TREATMENT PROGRAM EMPHASIZING THE ACQUISITION OF EMPLOYMENT-RELATED SKILLS. POST PROGRAM OUTCOMES WERE EVALUATED FOR THESE CLIENTS BY ASSESSING THEIR ADJUSTMENT IN TERMS OF RECIDIVISM AND EMPLOYMENT WHILE THEY REMAINED UNDER OFFICIAL SUPERVISION. THE GROUP WAS EQUALLY DIVIDED BETWEEN RECIDIVISTS AND NONRECIDIVISTS. NO SIGNIFICANT DIFFERENCES IN RECIDIVISM BY SEX OR RACE WERE FOUND. WHITES TENDED TO BE EMPLOYED OR INVOLVED IN SCHOOL OR TRAINING FOR LONGER PERIODS AFTER THEIR RELEASE FROM THE HALFWAY HOUSES THAN DID BLACKS. NO SUCH DIFFERENCES WERE OBSERVED FOR MALES AND FEMALES. THE FINDINGS POINT TO A RELATIONSHIP BETWEEN EMPLOYMENT AND RECIDIVISM. LONGER PERIODS OF EMPLOYMENT CHARACTERIZED CLIENTS WITH LOWER RECIDIVISM RATES, BUT THE RELATIONSHIP OF UNEMPLOYMENT TO RECIDIVISM WAS UNCLEAR. AN EFFORT TO DETERMINE WHETHER POST-PROGRAM SUCCESS COULD BE PREDICTED FROM EMPLOYMENT-RELATED VARIABLES AND UNEMPLOYMENT FOUND THAT, ON ALL CRITERION VARIABLES FOR WHICH SUCH PREDICTION WAS POSSIBLE, THE STRENGTH OF PREDICTORS VARIED CONSIDERABLY AMONG SUBGROUPS (BLACK, WHITES, MALES, FEMALES). SUPPORTING DATA, A REVIEW OF RELATED RESEARCH, AND A LIST OF REFERENCES ARE INCLUDED. (LKM).