We exploit the 1983 language-in-education reform that introduced Catalan alongside Spanish as medium of instruction in Catalan schools to estimate the labour market value of bilingual education. Identification is achieved in a difference-in-differences framework exploiting variation in exposure to the reform across years of schooling and years of birth. We find positive wage returns to bilingual education and no effects on employment, hours of work or occupation. Results are robust to education-cohort specific trends or selection into schooling and are mainly stemming from exposure at compulsory education. We show that the effect worked through increased Catalan proficiency for Spanish speakers and that there were also positive effects for Catalan speakers from families with low education. These findings are consistent with human capital effects rather than with more efficient job search or reduced discrimination. Exploiting the heterogeneous effects of the reform as an instrument for proficiency we find sizeable earnings effects of skills in Catalan.