Clinical reasoning with generation and prioritization of differential diagnoses (DDx) is a key skill for medical students, but no consensus exists on the best method to teach these skills. Meta-memory techniques (MMTs) may be useful, but the efficacy of individual MMTs is unclear. We designed a 3-part curriculum for pediatric clerkship students to teach one of 3 MMTs and provide practice in DDx generation through case-based sessions. Students submitted DDx lists during two sessions and completed pre- and post-curriculum surveys assessing self-reported confidence and perceived helpfulness of the curriculum. Results were analyzed using ANOVA with multiple linear regression. A total of 130 students participated in the curriculum, with 96 % (125/130) completing at least one DDx session, and 44 % (57/130) completing the post-curriculum survey. On average, 66 % of students rated all three sessions as "quite helpful" (4/5 on 5-point Likert scale) or "extremely helpful" (5/5) without difference between MMT groups. Students generated an average of 8.8, 7.1 and 6.4 diagnoses using the VINDICATES, Mental CT, and Constellations methods, respectively. When controlling for case, case order, and number of prior rotations, students using VINDICATES produced 2.8 more diagnoses than those using Constellations (95 % CI [1.1,4.5], p<0.001). There was no significant difference between VINDICATES and Mental CT (Δ=1.6, 95 % CI [-0.2,3.4], p=0.11) or Mental CT and Constellations (Δ=1.2, 95 % CI [-0.7,3.1], p=0.36). Medical education should include curricula focused on enhancing DDx development. Although VINDICATES helped students produce the most DDx, further research is needed to identify which MMT generates more accurate DDx.