Educators are increasingly using peer tutoring procedures in their work with underachieving students. Anecdotal reports suggest that peer tutoring may produce both academic gains and positive attitude changes in the tutor, including heightened self-esteem and increased motivation to learn (Gartner, Kohler, & Riessman, 1971). Systematic evaluations of peer-tutoring programs provide some support for such gains but also raise questions about the generality of these effects. Although positive gains in achievement for tutors are reported by Rust (1970), Liette (1972), and Allen and Feldman (1973), other investigators have found no specific achievement effects (Erikson & Cromack, 1972; Rogers, 1970). Similarly, gains in attitudes have been reported by some investigators (Haggerty, 1971) but not by others (Foster, 1972; Horan, DeGirolomo, Hill, & Shute, 1974). These empirical investigations of peer tutoring have left several questions unanswered. Only limited attention has been directed toward the attitudes of both tutor and tutee. Additionally, although not all studies have found measurable effects, the causes of success or failure of any one program are unclear. Apparently, peer tutoring is effective under some conditions but not under others. Applied peer-tutoring programs often differ in many ways, making it difficult to isolate the factors that may be mediating attitude change. Outcome has been related to a variety of program characteristics, ranging from tutor sex (Cicirelli, 1972), race (Lakin, 1972), and previous training (Niedermeyer, 1970) to the amount of tutee imitation (Allen & Devin-Sheehan, Note 1) or the effects of tutor–tutee sibling relationships (Cicirelli, 1972). It remains unclear, however, how these diverse factors may mediate attitude change. Without a conceptual model of the attitude change processes involved in peer tutoring, it is difficult to integrate such findings. Recently, role theory has been suggested as a conceptual framework within which peer tutoring effects may be explored and analyzed (Allen, 1976). According to role theory, enactment of a role produces changes in behavior, attitudes, and self-perceptions consistent with role expectations. Allen and Feldman (1973) suggested that the role of teacher represents competence, prestige, and authority. A child playing the role of teacher in a peer-tutoring situation may be expected to show changes in attitudes along these dimensions. As tutors begin to perceive themselves as more competent, they should develop more positive attitudes toward school and learning. The prestige and authority of the tutor role should also increase their feelings of importance, power, and self-worth. Children’s perceptions of their roles may be a central determinant of the attitude changes produced during tutoring experiences. Theoretically, the more that situational variables increase children’s perceptions of themselves as teachers, the greater the impact their role enactment will have. For example, when situations are structured to maximize children’s perceptions of the tutor role as a reflection of their own competence, greater positive changes in their self-concept may result. One way in which contextual variables may exert influence on role perceptions is through the initial role assignment process. Different procedures of role assignment may generate perceptions of tutor roles that vary in degree and quality of similarity to teacher roles. For example, when children are told that they have been assigned the role of tutor because of demonstrated competence, perceived similarity to a teacher is high and based on an achieved and valued quality. When role assignment is based on a teacherlike physical characteristic, perceived similarity is also high but on a less important dimension. Similarity between the teacher role and the tutor role may be perceived as low when no teacher qualities are necessary for assignment to tutor, as when role assignment is based on chance. The effects of contextual variables on the attitudes of tutees are less clearly defined in a role-theory analysis. To some extent, however, one would expect that conditions bolstering the role of the peer tutor would reflect negatively on the status of the tutee role. The purpose of this study was to evaluate a role-theory analysis of the attitudinal effects of peer tutoring on tutors and tutees. In particular, we hypothesized that attitude changes resulting from tutoring experiences are mediated by role perceptions and that contextual variables affect these role perceptions. To evaluate these hypotheses, fourth-grade children were assigned to be tutors or tutees, and they engaged in brief tutoring interactions. The assigned role was expected to have a major effect on role perceptions, and children enacting tutor roles were therefore expected to have more positive attitudes than children enacting tutee roles. The effects of contextual variables on role perceptions and attitudes were evaluated by providing children with four different rationales of role assignment. Some children were told that their role assignment was based on their prior performance (competence), some were told that assignment was based on a teacherlike physical quality (a clear voice), and some were told that assignment was based on chance (a coin toss). In the fourth condition, children received no information about the basis for their role assignment. We expected that role enactment would produce the most substantial effects for tutors when role assignment was based on competence. We predicted that tutoring would be somewhat less effective when role assignment was based on chance. When no basis for assignment was given, we expected that the tutor’s role perceptions would reflect the societal role of teacher and, hence, would produce positive attitudinal effects, similar to the role perceptions of children who were assigned on the basis of competence. The effects of the three rationales of role assignment on tutees’ attitudes were expected to be the reverse of the effects on tutors’ attitudes. When no rationale for role assignment was given, tutees were not expected to assume any deficiency in themselves relative to the tutor and were therefore expected to have attitudes similar to children assigned to tutees on the basis of chance. Additionally, the effects of the tutoring experience on different dimensions of attitudes were assessed by including measures of self-evaluation, attitudes toward the role, and attitudes toward the task. It was predicted that the effects of role and assignment condition would be strongest on attitudes toward one’s role, since role perceptions were expected to play a central function in mediating other attitudinal effects.