THE LONG RANGE PURPOSE OF PAROLE PREDICTION RESEARCH IS OUTLINED, WITH PARTICULAR ATTENTION TO PREVIOUS RESEARCH EFFORTS. A LITERATURE REVIEW IS PRESENTED, EMPHASIZING PREINSTITUTIONAL FACTORS, INTRA-INSTITUTIONAL FACTORS, POSTINSTITUTIONAL FACTORS, AND THE HYPOTHESES FOR THE PRESENT STUDY. THE RESEARCH METHODOLOGY IS THEN DETAILED. TWO TYPES OF MULTIVARIATE ANALYSIS WERE USED, AND 2 GROUPS OF SUBJECTS, 37 PAROLE RECIDIVISTS AND 101 PAROLE NONRECIDIVISTS, WERE SELECTED. USING INFORMATION OBTAINED FROM THE RECORDS OF THE GEORGIA TRAINING AND DEVELOPMENT CENTER AND THE GEORGIA STATE BOARD OF PARDONS AND PAROLE, NINE VARIABLES WERE COMPUTED IN A SIMPLE CORRELATION WITH A RECIDIVISM CRITERION IN ORDER TO TEST THE HYPOTHESIS. THE VARIABLES INCLUDE: (1) NUMBER OF PRIOR FELONY CONVICTIONS, (2) TYPE OF CRIME, (3) PARTICIPATION IN VOCATIONAL TRAINING, (4) PARTICIPATION IN ACADEMIC PROGRAMS, (5) PROCESSING BY THE DIAGNOSTIC CENTER, (6) AGE AT TIME OF PAROLE, (7) HOURLY WAGE ON PAROLE JOB, (8) NUMBER OF CONTACTS WITH PAROLE OFFICER DURING FIRST 3 MONTHS, AND (9) PAROLE OFFICE TO WHICH ASSIGNED. THE NINE FACTORS WERE INVESTIGATED TO TEST THE PREDICTION THAT THERE WOULD BE A SIGNIFICANT LINEAR RELATIONSHIP BETWEEN THE INDEPENDENT VARIABLES AND THE CRITERION. IN ADDITION, A DISCRIMINANT ANALYSIS WITH TWO GROUPS WAS COMPUTED TO OBTAIN A RANKING OF ALL SUBJECTS IN TERMS OF THEIR DISCRIMINANT SCORES AND TO DETERMINE THE OPTIMUM VALUE WHICH DISCRIMINATED RECIDIVISTS FROM NONRECIDIVISTS. THE RESULTS OF THE ANALYSES REGARDING THE THREE FACTORS IDENTIFIED DURING THE LITERATURE REVIEW AND THE PREDICTED CORRELATIONS BETWEEN INDEPENDENT VARIABLES ARE PRESENTED. A MAJOR FINDING OF THIS STUDY WAS THAT 7 OF THE 12 PREDICTED CORRELATIONS WERE STATISTICALLY SIGNIFICANT. THERE WERE SIGNIFICANT POSITIVE CORRELATIONS BETWEEN THE CRITERION AND THE VARIABLES OF HOURLY WAGE, NUMBER OF CONTACTS, AND PROCESSING BY THE DIAGNOSTIC CENTER. A SIGNIFICANT NEGATIVE CORRELATION WAS FOUND BETWEEN THE CRITERION AND THE NUMBER OF PRIOR FELONY CONVICTIONS. USING DISCRIMINANT ANALYSIS, 70 PERCENT OF THE RECIDIVISTS AND 80 PERCENT OF THE NONRECIDIVISTS WERE CLASSIFIED CORRECTLY. THE RESULTS ARE DISCUSSED IN DETAIL, AND A SUMMARY OF THE STUDY EFFORT IS PROVIDED. NOTES AND REFERENCES ARE INCLUDED. THE RANKING OF THE DISCRIMINANT SCORES FROM A DISCRIMINANT ANALYSIS WITH TWO GROUPS AND GROUP DIFFERENCES IN DISCRIMINANT SCORES FROM A DISCRIMINANT ANALYSIS WITH TWO GROUPS ARE APPENDED. TABULAR DATA ARE PRESENTED. (KBL)