CLASSES WERE DIVIDED EXPERIMENTALLY INTO FIVE OR SIX SMALL GROUPS. EACH GROUP WORKED TOGETHER FOR A PERIOD OF 1 MONTH. EACH OF THE EIGHT PARTICIPATING TEACHERS SELECTED ONE TO SIX PUPILS IN HIS CLASS WHO WERE EITHER DISRUPTIVE OR WITHDRAWN. THESE CHILDREN WERE CALLED THE TARGET PUPILS. THREE TIMES EVERY MONTH THE TEACHER RECORDED THE INTERACTION AND PLANNED FOR IMPROVEMENT. FOUR TYPES OF GROUPS WERE FOUND--THE LEADER DIRECTED GROUP, THE BALANCED-INTERACTIONGROUP, THE GROUP WHICH INTERACTED ONLY MINIMALLY, AND THE GROUP DOMINATED BY A MEMBER OTHER THAN THE DESIGNATED LEADER. ABOUT HALF OF THE SCHOOL DAY WAS SPENT IN SMALL GROUPS. ALL PUPILS, INCLUDING TARGET PUPILS, RESPONDED WELL TO THE PROGRAM. TEACHERS, THOUGH BOTHERED BY THE NOISE LEVEL, RESPONDED WELL TO TEACHING IN SMALL GROUPS. THE BALANCED TYPE OF SMALL GROUP PREDOMINATED. TARGET PUPILS DEVELOPED BETTER SOCIAL ADJUSTMENT AND LEADERSHIP CAPACITY. ON THE WHOLE, THE TRAINING OF THE TEACHER AS A DIRECTOR OF SMALL GROUPS AND THE USE OF SOCIOMETRICALLY FORMED SMALL GROUPS IN THE CLASSROOM PRODUCED ENOUGH POSITIVE OVER-ALL EDUCATIONAL RESULTS TO ENCOURAGE THE CONTINUED USE OF THIS TRAINING APPROACH.