THE STUDY WAS CONCERNED WITH THE RATE OF INMATES JUDGED ADMINISTRATIVELY TO REQUIRE CANCELLATION OF PAROLE SUPERVISION AND RETURN TO PRISON TO FINISH THEIR TERMS. ONE IMPORTANT PROPERTY OF PAROLE RECIDIVISM IS THE PAROLE AGENT'S DESIGNATION OF THE PAROLEE AS A SUCCESS OR FAILURE. THUS, RECIDIVISM RATES DO NOT DEPEND ENTIRELY ON THE PAROLEE'S PERFORMANCE BUT ARE ALSO A FUNCTION OF THE AGENTS' JUDGMENTS. TWO PAROLE DISTRICT OFFICES IN CALIFORNIA WERE SELECTED FOR THE STUDY BECAUSE THEY DEMONSTRATED DIFFERENT RATES OF RECIDIVISM. SIXTEEN CASELOAD SUPERVISORS WERE ASKED TO JUDGE WHETHER THE PAROLEE SHOULD CONTINUE ON PAROLE OR RETURN TO PRISON ON THE BASIS OF INFORMATON CONTAINED IN NINE CASE HISTORIES. RESEARCH INSTRUMENTS ALSO WERE ADMINISTERED TO THE SUPERVISORS TO EVALUATE OFFICE ORIENTATIONS, TO MEASURE JOB SATISFACTION, AND TO COLLECT SOCIOMETRIC INFORMATION CONCERNING RANK ORDER OF AGENTS ACCORDING TO HOW HARD THEY WORK. THE STUDY FOUND THAT THE OFFICE HAVING JURISDICTION OVER THE PAROLEE MADE A DIFFERENCE; PAROLEES REPORTING TO AN OFFICE WITH A HIGH INCIDENCE OF RETURNING PAROLEES TO PRISON HAD A GREATER CHANCE OF HAVING PAROLE REVOKED. THE DATA SUGGESTED THAT A DIFFERENCE IN OFFICE ORIENTATION EXISTED; WITHIN EACH UNIT THERE WAS A RANGE OF AGENT BEHAVIOR, AND A DECISION CONCERNING PAROLEE'S PERFORMANCE WAS PARTIALLY A PRODUCT OF THIS SOCIAL ORGANIZATION OF THE DISTRICT OFFICE. IT IS NOT CLEAR, HOWEVER, WHETHER THIS OFFICE ORIENTATION WAS ESTABLISHED BY THE DISTRICT OFFICE SUPERVISOR OR THE OFFICE'S COMMUNITY SETTING. FINDINGS INDICATED THAT OFFICER'S JOB DISSATISFACTION TENDED TO BE ASSOCIATED WITH A MORE PERSONALIZED AND HUMANISTIC APPROACH AND A LOWER RATE OF RETURN. NO ASSOCIATION BETWEEN RANKINGS IN TERMS OF HARDEST WORKING AGENTS AND RETURNING PAROLEES TO PRISON WAS OBSERVED IN ONE DISTRICT OFFICE, BUT IN THE SECOND OFFICE, HARDEST WORKING AGENTS WERE ASSOCIATED WITH LOWER RETURN RATES. THESE FINDINGS INDICATE THAT THE EFFECTS OF AGENTS' JUDGMENTS ON THE RECIDIVISM RATES MAY CANCEL THE EFFECTS OF TREATMENT. (KJM)