Although the education-health relationship is well documented, pathways through which education influences health are not well understood. This study uses data from a 2003-2004 cross sectional supplemental survey of respondents to the longitudinal Health and Retirement Study (HRS) who had been diagnosed with diabetes mellitus to assess effects of education on health and mechanisms underlying the relationship. The supplemental survey provides rich detail on use of personal health care services (e.g., adherence to guidelines for diabetes care) and personal attributes which are plausibly largely time invariant and systematically related to years of schooling completed, including time preference, self-control, and self-confidence. Educational attainment, as measured by years of schooling completed, is systematically and positively related to time to onset of diabetes, and conditional on having been diagnosed with this disease on health outcomes, variables related to efficiency in health production, as well as use of diabetes specialists. However, the marginal effects of increasing educational attainment by a year are uniformly small. Accounting for other factors, including child health and child socioeconomic status which could affect years of schooling completed and adult health, adult cognition, income, and health insurance, and personal attributes from the supplemental survey, marginal effects of educational attainment tend to be lower than when these other factors are not included in the analysis, but they tend to remain statistically significant at conventional levels.