There is limited evidence on the labor market impact of diabetes, and existing evidence tends to be weakly identified. Making use of Mexican panel data to estimate individual fixed effects models, we find evidence for adverse effects of self-reported diabetes on employment probabilities, but not on wages or hours worked. Complementary biomarker information for a cross section indicates a large diabetes population unaware of the disease. When accounting for this, the negative relationship of self-reported diabetes with employment remains, but does not extend to those unaware. This difference cannot be explained by more severe diabetes among the self-reports, but rather worse general health.