Introduction: Mastery learning, which deconstructs a complex task into sequential sub-steps combined with deliberate practice to achieve each step in sequence, represents an important method to enhance simulation-based procedural skills training. However, the evidence to support the effectiveness of this theory to improve learning is lacking. This study compared mastery learning using deliberate practice with self-guided practice on skill performance of a rarely performed, life-saving procedure, a bougie-assisted cricothyroidotomy (BAC). Methods: In this multi-centre, randomized study at five North American emergency medicine (EM) residency training programs, we assigned 166 EM postgraduate trainees to either mastery learning and deliberate practice (ML + DP) or self-guided practice for BAC. Three blinded airway experts independently evaluated BAC skill performance by video review before (pre-test) and after (post-test) each training session. The primary outcome was post-test skill performance using a 5-point global rating score (GRS). A secondary outcome, defined a priori, was performance time to complete the BAC skill (chronometry). Results: There was no significant difference in post-test BAC performance after ML + DP or self-guided practice. Performance scores improved for both groups by 13% from the pre-test to post-test (F (1,138) = 43, p < 0.001). Overall, time to complete the BAC improved significantly from pre-test (87.6 seconds) to posttest (54.1 seconds), (F, 1,149) = 122, p < 0.001). At post-test, the ML + DP group performed the skill 7.4 seconds faster than the self-guided practice group (F (1,150) = 6.77, p < 0.01). Conclusion: Mastery learning coupled with deliberate practice provides systematic and focused feedback during skill acquisition. However, it is resource intensive and its efficacy is not fully defined. In this study, MP + DP did not result in improved global performance; it did result in faster performance times, a relevant finding for time-sensitive procedures. These results are important for educators who seek to optimize technical skills training in a competency-based model of medical education. Our findings suggest that time-sensitive procedures might benefit from ML + DP teaching strategies to enhance time to procedural performance.