Background: Although peer-assisted learning is known to be effective for reciprocal learning in medical education, it has been understudied in simulation. We aimed to assess the effectiveness of peer-led compared to instructor-led debriefing for non-technical skill development in simulated crisis scenarios. Methods: Sixty-one undergraduate medical students were randomized into the control group (instructor-led debriefing) or an intervention group (peer debriefer or peer debriefee group). After the pre-test simulation, the participants underwent two more simulation scenarios, each followed by a debriefing session. After the second debriefing session, the participants underwent an immediate post-test simulation on the same day and a retention post-test simulation two months later. Non-technical skills for the pre-test, immediate post-test, and retention tests were assessed by two blinded raters using the Ottawa Global Rating Scale (OGRS). Results: The participants’ non-technical skill performance significantly improved in all groups from the pre-test to the immediate post-test, with changes in the OGRS scores of 15.0 (95% CI [11.4, 18.7]) in the instructor-led group, 15.3 (11.5, 19.0) in the peer-debriefer group, and 17.6 (13.9, 21.4) in the peer-debriefee group. No significant differences in performance were found, after adjusting for the year of medical school training, among debriefing modalities (P = 0.147) or between the immediate post-test and retention test (P = 0.358). Conclusions Peer-led debriefing was as effective as instructor-led debriefing at improving undergraduate medical students’ non-technical skill performance in simulated crisis situations. Peer debriefers also improved their simulated clinical skills. The peer debriefing model is a feasible alternative to the traditional, costlier instructor model.