In this paper, we estimate the relative earnings returns for Swedish upper secondary schooling programs with vocational and general content for individuals born 1955-76. These data contain detailed information from applications about students' choices which allows us to address two important sources of endogeneity. First, we use GPA admission cutoffs to generate a random element in program assignment. Second, we control for students' first best and second best choice combinations to proxy for students' comparative advantages. Combining these two strategies and merging data with annual earnings information from 1978-2011, we estimate the marginal returns to vocational schooling relative to general schooling. For men, our results indicate that vocational schooling has an earnings advantage over general schooling up and until the age of 27, with estimates thereafter close to zero. For women, we observe a similar earnings advantage for vocational schooling up and until the age of 23.