Teaching technical skills through the use of guided discovery learning (GDL) is an ongoing topic of research. In this approach, learners practice and struggle before receiving formal instruction. This has shown promise in other domains of learning, yet in the realm of procedural skills, clarity is still needed. This study seeks to address these gaps by investigating efficacy and mechanisms relating to application for a GDL approach in teaching basic surgical skills. In 2018, young surgical trainees (N = 16) undertook a 6-week open surgical course applying the principles of GDL, each lesson beginning with a discovery phase before subsequent instruction and practice. A concurrent triangulation mixed-methods approach was used with direct observation and collection of semistructured interviews using a framework designed from productive failure literature. At the end of the course, all participants took a conceptual knowledge test and a performance-based skills test. Performance on the skills test was rated using global ratings and checklists. The GDL cohort outperformed the historical cohort on the written exam (F [1,65] = 4.96, P = .029, d = .62), as well as on the summative suturing test (F [1,65] = 6.23, P = .015, d = .68). Furthermore, 3 main themes that highlight the mechanisms and mediators of efficient GDL were: (1) building conceptual knowledge, (2) motivating self-regulated learning, and (3) the type of skill and psychological safety. GDL can be an efficient approach to teaching procedural skills. Implications for future research and curricular design are discussed.