CASE FILES OBTAINED FROM THE CANADIAN NATIONAL PAROLE BOARD WERE USED TO COLLECT DATA ON SELECTED VARIABLES; THESE DATA WERE CODED, FREQUENCY TABULATIONS WERE MADE, RELATIONSHIPS BETWEEN PAIRS OF VARIABLES WERE INVESTIGATED, VARIABLES WERE COMBINED IN SETS TO FORM EQUATIONS OF VARIOUS SORTS 'PREDICTING' THE GRANT OR REFUSAL OF PAROLE AND THE KEEPING OR LOSS OF THE PARTIAL LIBERTY THAT PAROLE CONFERS, AND RELATIONSHIPS BETWEEN THE OTHER THREE SETS OF ANALYSES WERE THEMSELVES ANALYZED. IT WAS FOUND THAT: (1) IN USING ITEMS OF INFORMATION TO DECIDE WHETHER TO GRANT OR REFUSE PAROLE, THE PAROLE BOARD IS MORE INFLUENCED BY THE SOURCE OF AN ITEM THAN BY THE TOPIC TO WHICH THE ITEM REFERS; (2) HALF THE VARIANCE IN THE REPLY TO THE QUESTION 'WILL THIS APPLICANT BE GRANTED PAROLE OR NOT?' COULD BE EXPLAINED IN TERMS OF INFORMATION ON THE FILE AND THE STUDY'S TECHNIQUES OF ANALYZING IT; AND (3) ONLY 10 PERCENT OF THE VARIANCE IN THE REPLY TO THE QUESTION 'WILL THIS MAN KEEP OR LOSE HIS PARTIAL FREEDOM?' CAN BE EXPLAINED BY THE INFORMATION AVAILABLE AND THE TECHNIQUES OF ANALYZING IT, I.E., THE BEST PREDICTIOR OF WHETHER A PAROLEE'S FREEDOM WILL BE TAKEN AWAY IN THE FUTURE IS IF IT HAS BEEN TAKEN AWAY IN THE PAST. IT IS CONCLUDED THAT THE PAROLE BOARD'S CRITERIA FOR GRANTING OR REFUSING PAROLE ARE IRRELEVANT TO THE AIMS IMPLIED BY SUCH AN ACTIVITY, IF IT HAS AIMS AT ALL OTHER THAN PURELY BUREAUCRATIC ONES. BACKGROUND OF THE STUDY, RESEARCH METHODOLOGY, DATA ANALYSIS, AND FURTHER QUESTIONS SUGGESTED BY THE FINDINGS ARE DISCUSSED IN DETAIL. SUPPORTING DOCUMENTS, INCLUDING THE PAROLE ACT OF 1958, ARE APPENDED, AND REFERENCES ARE PROVIDED. STATISTICAL DATA ARE PRESENTED IN 117 TABLES, AND EQUATIONS ARE INCLUDED.