Background: To explore whether simulation-based endovascular training with focus on radiation safety could improve correct behavior without jeopardizing the learning of procedural skills. Methods: Twenty-four residents without previous endovascular experience completed 10 clinical scenarios on a virtual-reality endovascular simulator with software for peripheral endovascular interventions. Participants were randomized to receive feedback (n = 12) or not (n = 12) on radiation protection (RP) performance after each case. Expert assessments were done at the first, second, fourth, seventh, and 10th case on RP and endovascular skills (ES). Automatic simulator metrics on procedure time, contrast dose, handling errors, and estimated radiation exposure to patient and operator were registered. Outcome metrics were analyzed by two-way mixed analysis of variance pairwise comparisons with independent t-tests. Correlations were explored using Pearson's r for internal consistency reliability. Results: The RP performance was similar in both groups at their first attempt (P = 0.61), but the feedback group significantly outperformed the control group over time (P < 0.001 for all comparisons). The feedback group was however slower to learn the ES at start (P = 0.047 at second performance), but after 7 attempts no difference was shown (P = 0.59). The feedback group used more time (19.5 vs. 15.3 min; P = 0.007) but less contrast (60 vs. 100 mL; P < 0.001). The number of errors was the same in both groups, but all metrics regarding radiation exposure favored the feedback group (P-values from 0.001 to 0.008). Conclusions: Simulation-based training (SBT) is effective to acquire basic endovascular intervention skills and concurrently learn RP behavior when feedback on radiation culture is provided.