The vast majority of Mexico's indigenous students attend intercultural bilingual education (IBE) schools. They consistently underperform relative to their non-indigenous peers. The paper uses new data from the state of Chiapas to estimate the determinants of the achievement gap. Results from regression analysis with and without school fixed effects find weak evidence to suggest that IBE is associated with higher test scores in Mathematics for indigenous students when the model is implemented as intended. A descriptive analysis of the data finds that a large proportion of IBE schools in Mexico do not deliver a “fully-implemented” model where teachers are indigenous and speak the same language as their students. Further, results from the national competition for new teaching posts suggests that the supply of new, qualified IBE teachers is small. Taken together these results highlight the difficulties of IBE in Mexico to recruit qualified teachers who meet language requirements. IBE has many virtues and could have the potential to narrow the achievement gap. However, if policymakers do not make substantial efforts to develop a suitable teacher pipeline to staff IBE schools, the model's potential will be limited.