Two experiments are reported concerning the effects of the differential use of verbal approval by problematic adolescents serving as tutors in a remedial reading program for an inner‐city school. The experiments, each with 3 tutors and 15 tutees, used a combined multiple baseline and ABCBC design. Data showed that tutors' approvals as well as tutors' and tutees' on‐task and reading responses were low and stable during baseline. Tutors were trained to use verbal approval for tutees' on‐task behavior. Tokens were presented and withdrawn to control the tutors' use of approval. During phases in which tutors' approvals were raised via token dispensation, tutor reading and on‐task scores increased in a nonexperimental setting. Tutee reading scores also increased as a function of tutor approvals. The second experiment replicated these findings and, in addition, (a) tested the validity of changes in reading responses via standardized tests, (b) isolated and compared the covariance between variables in all phases, and (c) provided data on tutee attention to tutors as a possible natural reinforcer for the short‐term maintenance found in both studies. Data are discussed as evidence that tutors had acquired the ability to recruit reinforcement from the classroom for appropriate