Forty randomly selected, nonvolunteer, low achieving ninth graders selected from four classes in an inner city junior high school, were assigned the role of tutor to elementary school children in order to determine if this practice was promising as a method of compensatory education. In three of the four classrooms, the tutors and nontutors received the initial instruction together for three days. The instruction covered eleven objectives in fractions and the addition of fractions. For the next three weeks, tutors spent their math period tutoring fourth graders at a nearby elementary school, while nontutors worked in class practicing the same work that the tutors were teaching. These students formed a competing-treatment control group. Tutors from the remaining classroom were pulled out of class for the initial instruction, and students left in this fourth class continued to receive their regular math curriculum. They did not study the eleven objectives, and these students formed the no-treatment control group. Among the conclusions of the study drawn from results observed are the following: (1) this role change intervention was qualitatively different from other approaches that had been tried; (2) having to teach did produce learning in the teacher; (3) this intervention faced up to the responsibility of the schools to teach basic skills; and (4) this intervention affected secondary school children and elementary school children simultaneously. The study suggests that, initially at least, tutoring should be an assigned rather than a voluntary activity. (Author/AM)