PREVIOUS STUDIES HAVE SHOWN THAT REVOCATION DECISIONS AND HENCE, SUCCESS OR FAILURE RATES FOR PAROLEES, ARE INFLUENCED BY AT LEAST THREE FACTORS IN ADDITION TO THE BEHAVIOR OF THE PAROLEE: (1) COMMUNITY PRESSURES; (2) AGENCY PHILOSOPHY; AND (3) AGENCY POLICY. THIS PAPER EXAMINES THE ROLE OF AGENCY POLICY IN AN UNNAMED COMMUNITY. FOUR POLICIES WERE FOUND WHICH SEEMED TO SIGNIFICANTLY ALTER REVOCATION RATES. THE FIRST WAS THE HIRING OF PAROLE OFFICERS WITH A REHABILITATION, RATHER THAN A CONTROL, ORIENTATION. COLLEGE GRADUATES WITH A SOCIAL SCIENCES OR SOCIAL WORK BACKGROUND WERE SOUGHT. POLICE OR MILITARY EXPERIENCE BECAME LESS IMPORTANT. A SECOND POLICY AFFECTING SUCCESS RATES WAS THE OFFICIAL DESIGNATION OF THE AGENTS' TASK AS REHABILITATION AND THE UNSPOKEN IMPLICATION THAT 'GOOD' AGENTS HAD LOW REVOCATION RATES. AN UNINTENDED CONSEQUENCE OF THIS WAS THAT AGENTS OFTEN ACTED TO MITIGATE DAMAGES FROM CLIENT ACTS, OFTEN ARRANGING FOR RESTITUTION AND APOLOGIZING FOR THE CLIENT, ALL IN AN ATTEMPT TO AVOID PROSECUTION. A THIRD, CLOSELY RELATED POLICY, ENCOURAGED AGENTS TO DRAW ON COMMUNITY RESOURCES AS MUCH AS POSSIBLE. IF A CLIENT 'FAILED' WITH ONE AGENCY, HE WAS PASSED ON TO ANOTHER. A FOURTH POLICY WAS A DETERMINED ATTEMPT TO REDUCE REVOCATION RATES. WHILE THE CHANGE IN OFFICER ATTITUDE AND THE GREATER USE OF COMMUNITY AGENCIES MAY HAVE PROVED BENEFICIAL TO PAROLEES, THE DETERMINED EFFORT TO REDUCE REVOCATION RATES AND THE IMPLICATION THAT THE PAROLE AGENT HAD 'FAILED' IF RATES WERE NOT REDUCED LED TO CONSIDERABLE MANIPULATION OF REVOCATION FIGURES WITHOUT ANY REAL CHANGE IN OVERALL CLIENT BEHAVIOR. TO A PERSON TAKING THE RATES AT FACE VALUE, THE SYSTEM MAY APPEAR TO BE MORE SUCCESSFUL THAN THE OLD SYSTEM; HOWEVER, ONE NEEDS TO ASK HOW THE RATES WERE ACHIEVED. FINDINGS OF THE STUDY ARE DESCRIBED IN DETAIL. (GLR)