A TOTAL OF 1,493 JUVENILES WERE DRAWN RANDOMLY FROM THOSE PLACED ON PROBATION IN JAPAN FROM DECEMBER 1971, TO FEBRUARY 1972, AND THEIR BACKGROUNDS, APPLICATION OF THE CLASSIFICATION TABLE, AND RECIDIVISM WERE INVESTIGATED AS OF NOVEMBER 10, 1973. THE LENGTH OF THE FOLLOWUP PERIOD LASTED 2 YEARS. THE OUTCOME OF THE FOLLOWUP STUDY WAS AS FOLLOWS: (1) NONFAILURE GROUP (NO ABSCONDING AND NO ARREST) -- 53.2 PERCENT; (2) PETTY FAILURE GROUP (ABSCONDING, DISMISSAL WITH OR WITHOUT HEARING FOR A FRESH OFFENSE) -- 31.8 PERCENT; (3) MAJOR FAILURE GROUP (FRESH PROBATION ORDER, TRAINING SCHOOL ORDER, SUSPENSION OF PROSECUTION, FINE, IMPRISONMENT) --14.9 PERCENT. WHILE THE POINTS GIVEN BY THE EXISTING CLASSIFICATION TABLE IN THE BEGINNING OF PROBATION SHOWED AN AGREEABLE EFFECTIVENESS IN PREDICTING LEVELS OF ADJUSTMENT EVALUATED BY PROBATION OFFICERS 2 YEARS LATER, SUCH LEVELS ARE DEEMED TOO SUBJECTIVE TO QUALIFY FOR PRECISE CRITERION OF PREDICTION. VARIOUS ATTEMPTS AT REVISION OF THE CLASSIFICATION TABLE LED TO THE CONCLUSION THAT THUS FAR A NEW 14-ITEM TABLE PRODUCED BY THE MULTIPLE CORRELATION RATIO METHOD IS MOST EFFICIENT AND QUALIFIED TO SUBSTITUTE FOR THE PRESENT TABLE IN PREDICTING THE CONDUCT OF PROBATIONERS AFTER TWO YEARS. A TABLE IS INCLUDED TO SHOW THE PREDICTING POWERS OF THE VARIOUS METHODS TESTED. (RCB)