We investigate the workforce heterogeneity of startups with respect to education, age and wages. Our explorative study uses data on the population of 1,614 Danish firms founded in 1998. We track these firms until 2001 which enables us to analyze changes in workforce composition over time. Such a dynamic analysis constitutes a hitherto neglected area of entrepreneurship research. To assess relative workforce heterogeneity, we construct a simulated benchmark to which we compare observed workforce heterogeneity. We find that the initial workforce is relatively homogeneous compared to our benchmark. Our result holds both for non-knowledge-based and, to a lesser extent, knowledge-based startups. This seems surprising since a vast management literature advocates heterogeneous teams. The difficulties associated with workforce heterogeneity (like affective confl ict or coordination cost) as well as "homophily" (peoples inclination to bound with others with similar characteristics) hence appear to generally overweigh the benefits of heterogeneity (like greater variety in perspectives or more creativity). We also document that workforces become more heterogeneous over time - startups add workers with skills different from the workforce at startup. The initial supposedly "poor" mix of workforce characteristics is hence adjusted as the startup matures. This increase in workforce heterogeneity is, however, smaller compared to our benchmark but substantially larger than is team additions had the same characteristics as the initial team members.