We evaluated the effects of teaching 13 typically developing children to use constant time delay when tutoring three of their peers who had substantial disabilities. We measured the extent to which the tutors used the constant time delay procedure correctly and the tutees' correct performance on the skills being taught. A multiple probe design across behaviors was used. The results indicated that (a) the tutors used each step of the constant time delay with a high degree of procedural fidelity; (b) the students with disabilities acquired the skills that were taught during peer tutoring; and (c) modifications of the procedure were required for 2 of the 3 students with disabilities, and the peers tutors were able to implement the modifications. Implications for practice and future research are described.