OBJECTIVE Although several inanimate bench models have been described for training of suture skills, so far, there is no ideal method for teaching and learning this skill during medical education. The aim was to evaluate whether bench model fidelity interferes in the acquisition of suture skills by novice medical students. METHODS 36 medical students with no surgical skills' background (novices) were randomized to three groups (n = 12): theoretical suture training alone (control); low-fidelity suture training model (synthetic ethylene-vinyl acetate bench model); or high-fidelity suture training model (pig feet skin bench model). Pre- and post-tests were applied (performance of simple interrupted sutures and subdermal interrupted sutures on ox tongue). Three tools (Global Rating Scale with blinded assessment, effect size, and self-perceived confidence based on Likert scale) were used to measure all suture performances. RESULTS The post-training analysis showed that the students that practiced on bench models (hands-on training) presented better (all p < 0.0000) performance in the Global Rating Scale evaluation, compared with the control, regardless of the model fidelity. The magnitude of the effect (training) was considered large (> 0.80) in all measurements. Students felt more confident (all p < 0.0000) to perform both types of sutures after training. CONCLUSION The acquisition of suture skills on the low-fidelity bench model was similar to that of the high-fidelity bench model, and the increase in the performance of participants that received bench model training was superior to those who received training based on theoretical teaching materials.