Medical schools are increasingly faced with a more diverse student population. Generally, ethnic minority students are reported to underperform compared with those from the ethnic majority. However, there are inconsistencies in findings in different types of examinations. Additionally, little is known about the performance of first-generation university students and about performance differences across ethnic minority groups. This study aimed to investigate underperformance across ethnic minority groups and by first-generation university students in different types of written tests and clinical skills examinations during pre-clinical training. A longitudinal prospective cohort study of progress on a 3-year Dutch Bachelor of Medicine course was conducted. Participants included 2432 students who entered the course over a consecutive 6-year period (2008–2013). Compared with Dutch students, the three non-Western ethnic minority groups (Turkish/Moroccan/African, Surinamese/Antillean and Asian) underperformed in the clinical problem solving tests, the language test and the OSCEs. Findings on the theoretical end-of-block tests and writing skills tests, and results for Western minority students were less consistent. Age, gender, pre-university grade point average and additional socio-demographic variables (including first-generation university student, first language, and medical doctor parent) could explain the ethnicity-related differences in theoretical examinations, but not in language, clinical and writing skills examinations. First-generation university students only underperformed in the language test. Apparently, underperformance differs both across ethnic subgroups and between different types of written and clinical examinations. Medical schools should ensure their assessment strategies create a level playing field for all students and explore reasons for underperformance in the clinical and writing skills examinations.