OFFENDERS WERE 646 FIRST INCARCERATES AT THE GUELPH CORRECTIONAL CENTRE WHO WERE SERVING SENTENCES OF LESS THAN 2 YEARS. MOST WERE CONVICTED ON THEFT-RELATED CHARGES. THE MEAN AGE WAS 20.01 YEARS. MMPI SCALES ONE THROUGH NINE WERE SCORED, AS WELL AS PANTON'S PRISON ADJUSTMENT SCALE, PAROLE VIOLATION SCALE, AND HABITUAL CRIMINAL SCALE. FOURTEEN OTHER EXPERIMENTAL SCALES WERE USED TO DEVELOP A TEST BATTERY FOR ASSESSMENT AND RESEARCH IN CORRECTIONAL SETTINGS. EACH INMATE WAS INDIVIDUALLY INTERVIEWED ON A PERSONAL INFORMATION FORM AND WAS THEN TESTED. TESTING OCCURRED WITHIN 2 WEEKS OF ADMISSION. TWO YEARS AFTER INMATES' RELEASE FROM THE INSTITUTION, FINGERPRINT RECORDS WERE OBTAINED TO EVALUATE RECIDIVISM. MMPI SINGLE-SCALE CORRELATIONS WITH RECONVICTION WERE UNIFORMLY LOW. THE PERCENTAGE OF CASES CORRECTLY CLASSIFIED AS RECONVICTED WAS 66.5 FOR 12 OTHER SCALES. THE FALSE ALARM RATE WAS 32.9 PERCENT. HIGHEST STANDARDIZED DISCRIMINANT-FUNCTION COEFFICIENTS CAME FROM REJECTION, SOCIALIZATION, AND ASSAULT SCALES. THE CAPABILITY OF PSYCHOLOGICAL TESTS TO PREDICT RECIDIVISM IS CONTROVERSIAL. IF A MAJOR CONCERN OF CORRECTIONAL CLINICIANS IS TO PREDICT OFFENDER BEHAVIOR AFTER RELEASE, STUDY FINDINGS DO NOT PROVIDE COMPELLING EVIDENCE FOR THE ROUTINE USE OF TRAIT-LIKE MEASURES SUCH AS THE MMPI FOR DIAGNOSTIC PURPOSES. SUPPORTING DATA AND REFERENCES ARE INCLUDED. (DEP)