DATA ON 226 CASES OF ADULT MALE OFFENDERS WHICH WERE HEARD BETWEEN JANUARY AND JUNE OF 1974 WERE ANALYZED IN RELATION TO FACTORS CONSIDERED IN RELEASE DECISIONS BY THE BOARD: PRIOR CRIMINAL AND PAROLE RECORD; INSTITUTIONAL ADJUSTMENT; PARTICIPATION IN EDUCATIONAL AND TREATMENT PROGRAMS; AND ADEQUACY OF PAROLE PLANS. WHEN THESE FACTORS WERE DIVIDED INTO 21 SUBSTANDARDS AND FURTHER ANALYZED, IT WAS FOUND THAT, FOR 9 OF THE SUBSTANDARDS, BOARD PERFORMANCE WAS RATED UNSATISFACTORY OR UNCERTAIN. TWENTY-THREE PERCENT OF VARIATION IN BOARD DECISIONS COULD NOT BE JUSTIFIED UNDER BOARD STANDARDS. RECOMMENDATIONS FOR IMPROVING BOARD CONFORMITY TO BOARD STANDARDS INCLUDE: REVISING THE STANDARDS TO AGREE WITH ACTUAL PRACTICE; CLASSING STANDARDS INTO THOSE OF MAJOR AND MINOR IMPORTANCE; SUPPLYING INMATES AND BOARD MEMBERS WITH COPIES OF THE REVISED STANDARDS; AND UNDERTAKING FURTHER STATISTICAL STUDIES TO MONITOR BOARD PERFORMANCE. THE STUDY ALSO INCLUDED A FOLLOWUP OF THE 128 OFFENDERS WHO WERE RELEASED TO PAROLE, IN WHICH FACTORS THAT PREDICTED SUCCESS OR FAILURE WERE IDENTIFIED. (NINETY-TWO, OR 72 PERCENT, WERE CONSIDERED 'SUCCESSES' AND THIRTY-SIX, OR 28 PERCENT, WERE CONSIDERED 'FAILURES.') THE BENEFITS OF USING STATISTICAL METHODS FOR PREDICTING OUTCOME WERE SHOWN TO BE SUBSTANTIAL; IT IS RECOMMENDED THAT THE BOARD DEVELOP STATISTICAL PREDICTORS AND CONSIDER IMMEDIATELY USING THE MOST RELIABLE OF THE PREDICTORS -- DENIAL OF PAROLE AT LAST HEARING. OTHER CHANGES IN BOARD ORGANIZATION AND PROCEDURES SUGGESTED ON THE BASIS OF THE STUDY INCLUDE: PERMITTING COUNSEL AT BOARD HEARINGS; CREATING A BETTER METHOD OF SELECTING BOARD MEMBERS; GIVING INMATES LIMITED ACCESS TO THEIR FILES PRIOR TO PAROLE HEARINGS; TRAINING BOARD MEMBERS MORE INTENSIVELY; OPENING PAROLE HEARINGS TO THE PUBLIC; AND CREATING A SYSTEM FOR ESTABLISHING PRECEDENTS. TABLES PRESENT SUPPORTING STATISTICS.