THE EXPERIMENTAL SUBJECTS, WHOSE MEAN AGE WAS 10.93 YEARS, HAD BEEN PLACED IN A PRIVATE INSTITUTION IN UPSTATE NEW YORK BECAUSE OF THEIR SERIOUS ACTING OUT, AGGRESSIVE, AND PREDELINQUENT BEHAVIOR. MOST WERE FROM LOWER SOCIOECONOMIC STATUS, MULTIPROBLEM FAMILIES. THE CONTROL SUBJECTS WERE FROM A MIDDLE-CLASS SUBURB IN UPSTATE NEW YORK. THE YOUTHS COMPLETED THE MATCHING FAMILIAR FIGURES TEST, WHICH MEASURES THE DEGREE TO WHICH ONE PAUSES TO EVALUATE THE RESULTS OF ONE'S COGNITIVE EFFORTS IN A SITUATION CONTAINING RESPONSE UNCERTAINTY. REFLECTIVE PERSONS TAKE MORE TIME TO SOLVE THE TEST'S MATCHING TASKS AND ALSO TEND TO MAKE FEWER ERRORS THAN DO IMPULSIVE PERSONS. CONTRARY TO EXPECTATIONS, PREDELINQUENTS WERE MORE REFLECTIVE THAN NONDELINQUENTS ON BOTH THE LATENCY (TIME TAKEN TO PRODUCE SOLUTIONS) AND ERROR DIMENSIONS OF THE TEST. THE UNANTICIPATED REFLECTIVENESS OF THE PREDELINQUENTS COULD HAVE BEEN THE RESULT OF INSTITUTIONALIZATION EFFECTS, OF THE RESOCIALIZATION PROGRAM IN WHICH THE YOUTHS WERE INVOLVED, OR OF THEIR WARINESS REGARDING THE USES TO WHICH RESEARCHERS WOULD PUT THEIR TEST SCORES. SUGGESTIONS FOR FURTHER STUDIES ALONG THESE LINES ARE OFFERED. NO TABULAR DATA ARE INCLUDED. A LIST OF REFERENCES IS PROVIDED. (LKM)