This study aims to analyze the quality of mid-semester exam questions in the form of descriptive questions, focusing on the aspects of validity, reliability, level of difficulty, and discriminating power. The method used is a descriptive quantitative approach, with documentation techniques for odd semester mid-term exam questions and student answer sheets. Data analysis was carried out using Anates software version 4. The results showed that two of the four questions had very significant validity and reliability with a coefficient of 0.65, which is included in the high category. In terms of level of difficulty, one question was classified as easy (74%), two questions were moderate (53–54%), and one question was very difficult (14%). The discriminating power showed that one question was categorized as sufficient and three questions were categorized as good to very good. These findings indicate that most of the questions analyzed have a quality that is suitable for use in learning evaluation. The implications of this study emphasize the importance of item analysis as a step to improve the quality of assessment in higher education, especially in designing questions that are valid, reliable, and able to measure student learning achievement fairly. In addition, this study can be used as a reference in training in preparing evaluation instruments based on empirical data for lecturers and educators.