This project investigated the effects of training for peer response in university freshman composition classes over the course of one 15-week semester. Eight sections of composition (total n = 169) participated. Students in the experimental group, composed of four sections, were trained via teacher-student conferences in which the teacher met students in groups of three to develop and practice strategies for peer response. Students in the control group, also four sections, received no systematic training aside from viewing a video example. The experimental and the control groups were compared with respect to the quantity and quality of feedback generated on peer writing as well as student interaction during peer response sessions. Analyses of data indicated that training students for peer response led to significantly more and significantly better-quality peer feedback and livelier discussion in the experimental group.