Assessment is an important aspect of medical education because it tests students' competence and motivates them to study. Various assessment methods, with and without images, are used in the study of anatomy. In this study, we investigated the use of extended matching questions (EMQs). To gain insight into the influence of images on the validity of test items, we focused on students' cognitive processes while they answered questions with and without images. Seventeen first-year medical students answered EMQs about gross anatomy, combined with either labeled images or answer lists, while thinking aloud. The participants' verbal reports were transcribed verbatim and then coded. Initial codes were based on a task analysis and were adapted into final codes during the coding process. Results showed that students used more cues from EMQs with images and visualized more often in EMQs with answer lists. Ready knowledge and verbal reasoning were used equally often in both conditions. In conclusion, EMQs with and without images elicit different results in this think aloud experiment, indicating different cognitive processes. They seem to measure different skills, making them valid for different testing purposes. The take-home message for anatomy teachers is that questions without images seem to test the quality of students' mental images while questions with images test their ability to interpret visual information. It makes sense to use both response formats in tests. Using images from clinical practice instead of anatomical drawings will help to improve test validity.