(1) Background: Augmented reality (AR) simulation may accelerate psychomotor skill acquisition in clinical education, but comparative evidence is scarce. This three-arm randomized controlled trial compared AR simulation, basic task-trainer simulation, and lecture-based instruction for urinary catheterization training. We hypothesized that AR would be associated with higher performance compared to the other two methods. (2) Methods: Primary outcomes included male and female catheterization skills assessed with checklists. Secondary outcomes included knowledge and confidence. One-way ANOVA with Tukey’s HSD post hoc tests constituted the primary analysis; Kruskal–Wallis tests and Bayesian ANOVA provided convergent evidence. (3) Results: A total of 176 trainees were assigned to AR simulation, basic simulator, or lecture-only control groups (N = 60, 58, and 58, respectively). AR simulation was associated with higher skill scores than both basic simulation and the control for male catheterization (p < 0.001, η² = 0.574) and female catheterization (p < 0.001, η² = 0.535), with large effect sizes (Cohen’s d = 3.11, AR vs. control). Knowledge showed no group difference (p = 0.11). Confidence moderately favored AR (d = 0.69 vs. Control). Bayesian analysis supported a high probability of AR outperforming Control. (4) Conclusions: AR simulation training was associated with superior catheterization skills compared to both basic simulation and lecture-based instruction, with large effect sizes observed across both frequentist and Bayesian analyses. Knowledge was consistent across groups, suggesting a possible ceiling effect.