PURPOSE: To assess radiology residents' diagnostic accuracy in interpreting thoracic emergency cases using the Wisdom in Diagnostic Imaging Simulation (WIDI SIM) platform and identify potential areas for educational improvement. METHODS: In this retrospective study, 761 radiology residents were assessed on five thoracic emergency cases using WIDI SIM over four years. Cases included bronchial disruption, septic emboli, ventricular perforation, pulmonary embolism, and a negative pulmonary CTA. Residents provided free-text interpretations, which were scored by faculty using a standardized point system. Scores and errors were analyzed using descriptive statistics and the Kruskal-Wallis test. RESULTS: Residents' performance varied across the five cases, with the highest average score on the negative pulmonary CTA (9.59) and the lowest on bronchial disruption (6.59). Observational errors were more common than interpretive errors. The Kruskal-Wallis test revealed significant differences in median scores across the cases (p <
0.0001), with pairwise comparisons showing significant differences in all but two comparisons. CONCLUSION: This study reveals significant variability in radiology residents' diagnostic accuracy in interpreting thoracic emergency cases, with a high prevalence of observational errors. Our observations emphasize the need for targeted educational strategies to address specific areas of weakness and improve diagnostic accuracy in this critical area of radiology practice.