THE SETTING FOR THE STUDY WAS AN UPPER MIDDLE-CLASS NEIGHBORHOOD IN THE AREA OF OMAHA, NEBR., AND STUDENTS PARTICIPATED WITH THE CONSENT OF THEIR PARENTS. STUDENTS COMPLETED A QUESTIONNAIRE CONTAINING, AMONG OTHER ITEMS, A NUMBER OF QUESTIONS PERTAINING TO DELINQUENT BEHAVIOR. AFTER A 3-WEEK PERIOD, THEY COMPLETED ANOTHER QUESTIONNAIRE IN WHICH SELF-REPORTED DELINQUENCY ITEMS WERE REPEATED. FOR EACH OF THE 28 ITEMS ON THE QUESTIONNAIRE, STUDENTS REPORTED THE TOTAL NUMBER OF TIMES THEY HAD ENGAGED IN THAT BEHAVIOR AND THE NUMBER OF TIMES THEY HAD ENGAGED IN THAT BEHAVIOR WITHIN THE PAST YEAR. OVERALL THE TEST-RETEST SCALE CORRELATIONS WERE 0.89 OR GREATER, DEPENDING ON THE TYPE OF DATA. ON THE AVERAGE, 78 PERCENT OF THE STUDENTS REPORTED EXACTLY THE SAME FREQUENCIES OF BEHAVIOR ON THE SECOND COMPLETION OF THE QUESTIONNAIRE THAT THEY HAD ON THE FIRST. IT IS CONCLUDED THAT SUBJECTS CAN PROVIDE RELIABLE INFORMATION ABOUT THEIR DELINQUENT BEHAVIOR AND THAT THE INFORMATION CAN BE INTERPRETED RELIABLY IN SEVERAL DIFFERENT WAYS. WHEN THE RESEARCHER'S INTEREST LIES IN THE FREQUENCY OF A PARTICULAR TYPE OF DELINQUENT BEHAVIOR, IT APPEARS THAT ASKING SUBJECTS TO REPORT ONLY ON THE FREQUENCY OF THE BEHAVIOR AND NOT ON THE FREQUENCY OF BEHAVIOR WITHIN A SPECIFIED TIME PERIOD MAY PRODUCE MORE RELIABLE INFORMATION. IT ALSO APPEARS THAT RATIONAL TRANSFORMATIONS OF DATA (E.G., CONVERSION TO LOGARITHMS) MAY ENHANCE MEASUREMENT RELIABILITY. THE ESTABLISHMENT OF THE RELIABILITY OF SELF-REPORTED DELINQUENCY DATA IS NECESSARY FOR THE ACCURATE INTERPRETATION OF FINDINGS, AND IT IS PARTICULARLY IMPORTANT IN THE COLLECTION OF LONGITUDINAL DATA WHERE CHANGES IN BEHAVIOR OVER TIME ARE OF PRIMARY CONCERN. FOOTNOTES AND SUPPORTING TABULAR DATA ARE PROVIDED. APPENDIXES CONTAIN TABLES AND A LIST OF THE SELF-REPORTED DELINQUENCY ITEMS. (DEP)