ALL INMATES OF A BORSTAL IN SCOTLAND WERE INCLUDED IN THE ORIGINAL SAMPLE, GIVING A FINAL SAMPLE OF 34 AFTER ELIMINATING 6 BOYS FOR WHOM DATA WERE INCOMPLETE OR UNRELIABLE. FOR COMPARISON, A SAMPLE OF BOYS ATTENDING CITY HIGH SCHOOLS OR WORKING ON A YOUTH EMPLOYMENT SCHEME WERE TESTED. THE CONTROL GROUP WAS MATCHED WITH THE INCARCERATED GROUP IN TERMS OF AGE, SOCIOECONOMIC BACKGROUND, AND URBAN RESIDENCE. EACH BOY WAS INTERVIEWED INDIVIDUALLY BY THREE DIFFERENT PERSONS. ONE PERSON ADMINISTERED A BRIEF TEST OF INTELLIGENCE AND A SELF-REPORT DELINQUENCY INVENTORY. ANOTHER PERSON ADMINISTERED REST'S DEFINING ISSUES TEST (DIT) OF MORAL REASONING, WHILE THE THIRD PERSON ADMINISTERED THE MEASURE OF OPERATIONAL THOUGHT. TO PROVIDE ADDITIONAL AND MORE DIFFERENTIATED INFORMATION ABOUT OFFENSES, FOUR PEOPLE INVOLVED IN THE JUVENILE SYSTEM WERE ASKED TO RATE CRIMINAL ITEMS AS SERIOUS OR TRIVIAL. THE BORSTAL GROUP OBTAINED SIGNIFICANTLY HIGHER MEAN SCORES ON SCALES OF SELF-REPORTED DELINQUENCY THAN THE CONTROL GROUP. MALE DELINQUENTS ADMITTED TO PROPORTIONATELY MORE SERIOUS OFFENSES. THE CONTROL GROUP WAS NOT SIGNIFICANTLY BETTER THAN THE BORSTAL GROUP IN OVERALL PERFORMANCE ON LOGICAL OPERATIONAL ABILITY TASKS. THERE WAS A DIFFERENCE BETWEEN THE TWO GROUPS IN MATURITY OF MORAL REASONING. THE CONTROL GROUP WAS SLIGHTLY BUT SIGNIFICANTLY MORE ADVANCED ON THE DIT THAN THE BORSTAL GROUP. IT IS CONCLUDED THAT THE EXTENT AND SERIOUSNESS OF INVOLVEMENT IN DELINQUENCY ARE UNRELATED TO MATURITY OF MORAL REASONING AND THAT THE UTILITY OF THE COGNITIVE DEVELOPMENTAL APPROACH TO MORAL SOCIALIZATION MAY BE MORE RESTRICTIVE THAN PREVIOUSLY CLAIMED. FURTHER STUDY IS RECOMMENDED TO PROVIDE FIRM SUPPORT FOR A POSSIBLE LINK BETWEEN MORAL REASONING AND INVOLVEMENT OR NONINVOLVEMENT IN SERIOUS AND PERSISTENT CRIMINAL ACTIVITY. SUPPORTING DATA AND REFERENCES ARE INCLUDED. (DEP)