AS PART OF A LARGER RESEARCH PROJECT ON 'CRIMINAL CAREERS OF HABITUAL FELONS,' THE RELIABILITY OF THE INFORMATION OBTAINED FROM THE CRIMINALS THROUGH PERSONAL INTERVIEWS WAS ASSESSED. FIRST THE OFFENDER'S CRIMINAL HISTORY WAS CHECKED. THEN DATA GATHERED BY TRAINED INTERVIEWERS WERE CROSS-MATCHED WITH THE OFFICIAL RECORD. THE VARIETY OF TECHNIQUES USED TO AID RECALL ARE DESCRIBED. DURING THE INTERVIEWS THE OFFENDERS REPORTED APPROXIMATELY 20 OFFENSES PER YEAR OF STREET TIME, ABOUT 10 TIMES AS MANY AS SHOWED ON THE ARREST SHEETS. BY MATCHING SELF-REPORTS TO OFFICIAL RECORDS, IT WAS CALCULATED THAT ONLY 2.4 PERCENT OF CRIMES RESULT IN ARREST AND 1.8 PERCENT IN CONVICTION. ABOUT 75 PERCENT OF THE ARREST WERE ACCURATELY RECALLED. ERRORS INVOLVED LESS SERIOUS OFFENSES. EVENTS MOST ACCURATELY RECALLED WERE ROBBERY AND RAPE, WHEREAS THOSE LEAST ACCURATELY REPORTED WERE GRAND LARCENY, AGGRAVATED ASSAULT, AND AUTO THEFT. ALSO, EVENTS WHICH TOOK PLACE CLOSEST TO TIME OF INTERVIEW WERE LEAST ACCURATELY RECALLED. IT IS SUGGESTED THAT THE OFFENDERS DESIRED TO LOOK REHABILITATED. THE FINDINGS ALSO SHOWED THAT THE ANSWERS WERE EQUALLY TRUTHFUL FOR BOTH MALE AND FEMALE INTERVIEWERS. TABLES PRESENT STUDY DATA. AN EXTENSIVE BIBLIOGRAPHY IS APPENDED. (GLR)