This article focuses on the explanation of cross-country variation in graduate overeducation using a multilevel analysis. We use data from representative surveys among graduates in 13 European countries and Japan, focusing on the first job and the present job 5 years after graduation. We find that cross-country differences in overeducation are explained by the quality and orientation (general versus specific) of the educational system/programme, the business cycle and the relative oversupply of highly skilled labour. Employment protection legislation has no impact on the incidence and persistence of overeducation.