share of higher educated workers in close peer groups on wages, using a rich data source linking workers to workplaces and specific occupations. Our empirical approach accounts for the endogenous sorting of workers into peer groups and workplaces, and at the same time avoids the reflection problem which is a challenge for analyses of peer effects. In our main specification we find statistically significant but economically small peer effects across all occupations. The magnitude of the effect differs across length and type of education, as well as across occupations and peer group- and workplace size.