EACH OF THE 421 INMATES WAS EVALUATED BY A CASEWORKER, WHO INTERVIEWED THE SUBJECT, REVIEWED HIS FILE, AND WROTE A PREPAROLE REPORT TO THE ADULT AUTHORITY BOARD. THESE REPORTS, THE RECORDS OF THE ADULT AUTHORITY BOARD (IN 62 PERCENT OF CASES REASONS FOR THE PAROLE DECISION WERE GIVEN), AND THE 1-YEAR FOLLOW-UP REPORTS PREPARED BY THE PAROLE OFFICERS WERE THEN EXAMINED AND COMPARED. THE CASEWORKERS FOCUSED MAINLY ON RECIDIVISM-RELATED VARIABLES AND THE INMATES' BEHAVIOR WHILE IN PRISON. THEY TENDED TO RECOMMEND PAROLE MORE OFTEN FOR SERIOUS OFFENDERS THAN FOR THOSE CONVICTED OF LESS SERIOUS OFFENSES, PROBABLY REFLECTING KNOWLEDGE THAT SERIOUS OFFENDERS ARE LESS LIKELY TO BE RECIDIVISTS. ON THE OTHER HAND, THE PAROLE BOARD FOCUSED ON OFFENSE AND PLACED LITTLE EMPHASIS ON BEHAVIOR WHILE IN PRISON. PAROLE OUTCOME WAS POSITIVELY CORRELATED WITH CASEWORKER EVALUATION (A CORRELATION COEFFICIENT OF .097) AND NEGATIVELY CORRELATED WITH PAROLE BOARD EVALUATION (A CORRELATION COEFFICIENT OF MINUS 0.006, WHICH WAS NOT SIGNIFICANT). THERE WAS ALSO A WEAK RELATIONSHIP BETWEEN VARIABLES PAROLE BOARD MEMBERS SAID THEY USED AND THEIR ACTUAL USE OF INFORMATION, AS REPORTED IN THE OFFICIAL RECORDS. THE FINDINGS ARE DISCUSSED IN TERMS OF SOCIAL ROLE DIFFERENCES BETWEEN THE TRAINED CASEWORKERS AND THE PAROLE BOARD MEMBERS, WHO ARE POLITICAL APPOINTEES WITH VARYING DEGREES OF EXPERIENCE IN CRIMINAL JUSTICE. IT IS SUGGESTED THAT PAROLE GUIDELINES ARE NEEDED TO LIMIT DISCRETION AND REDUCE CONFUSION AMONG DECISIONMAKERS ABOUT THE APPROPRIATE WEIGHTING OF VARIABLES IN PAROLE OUTCOME PREDICTION. REFERENCES ARE INCLUDED. (GLR)