Background: The written test of the Endocrinology and Metabolism specialist board exam (EMSBE) contained 100 multiple-choice questions (MCQs). The objectives of this study were to evaluate and improve the quality of the MCQs. Methods: The written test was formatted by considering the distributions of the degree of difficulty (DD) of each question marked by the expert who posed the question. The MCQs of the years 2016 and 2018 were used to analyze the correct rate (CR), the difficulty index (PI), and the discrimination index (DI) of each item. Associations between DD and CR, PI, or DI were evaluated. Results: The average CR, PI, and DI was 59.9 % ± 24.6 %, 0.58 ± 0.24, and 0.30 ± 0.27, respectively, in 2016 and 65.0% ± 25.8%, 0.60 ± 0.26, and 0.27 ± 0.32, respectively, in 2018. Defined by 0.35 ≤ PI < 0.60, there were 33 items and 24 items with "optimal" difficulty in 2016 and 2018, respectively. Defined by DI ≥ 0.40, there were 33 items and 35 items with "very good" discriminatory power in 2016 and 2018, respectively. The associations between DD and CR, PI, or DI were statistically non-significant, both in 2016 and 2018. Conclusions: Not all MCQ items used in the written test for EMSBE had optimal difficulty or sufficient discriminatory power. The DD marked by the expert who posed the question was not associated with CR, PI, or DI. This study reviewed the MCQ items of EMSBE and established the basis for future quality improvements.