Introduction The simulation of patient death remains controversial in simulation-based education. We investigated the effect of simulated patient death on learners' skill retention, stress levels, and emotions. Methods After ethics approval, we recruited residents at two Canadian universities. Participants were randomized to manage a simulated cardiac arrest ending with either the unexpected death (intervention group) or survival (control group) of the simulated patient (i.e., manikin). Three months later, all participants performed the same scenario but with the opposite outcome. Blinded video raters assessed participants' non-technical and technical crisis resource management (CRM) skills at both time points. Stress levels (represented by anxiety level, salivary cortisol concentration, and cognitive appraisal) and emotional valence were measured. Outcomes were analyzed using analysis of covariance (ANCOVA) or generalized estimating equations as appropriate. Results The analysis included 46 participants (intervention: n=24; control: n=22). Simulated death neither affected retention of non-technical CRM skills (mean retention Ottawa Global Rating Scale score in the death group [29.4, 95% CI: 27.0, 31.8] versus control group [29.4, 95% CI: 26.8, 32.0; p= 0.87]) nor technical CRM skills (mean retention task-specific checklist score in the manikin death group [11.8, 95% CI: 10.5, 13.0] versus the control group [12.5, 95% CI: 11.3, 13.7; p=0.69]). The simulated death had negative effects on participants' anxiety levels, cognitive appraisal, and emotions. Conclusion Simulated patient death did not affect the retention of non-technical or technical CRM skills but led to greater levels of short-term anxiety, stress, and negative emotions among learners.