The study proposed a mix-mode peer response, the E-Peer Response (EPR), to overcome the bias of a single mode, and examined how students with different levels of ability react to the EPR. Two classes participated in this study. One was the experimental group (EG) with the EPR; the other was the control group (CG) with a teacher-centered writing approach. Composition posttest scores were applied to evaluate participants' writing performance while the number of feedback was employed to evaluate students' peer response behavior. The results revealed that students in the EG performed better than those in the CG. This was due to the fact that the EPR provided a convenient online writing environment, a complete writing practice, and a sharable mechanism. High-ability students not only benefited more from the EPR than the low-ability students, but also high-ability students significantly outperformed low-ability students in giving suggestions on content features.