This study examines the education gradient in three chronic conditions - diabetes, hypertension, and high cholesterol. In the analysis, we take into account diagnosed as well as undiagnosed cases, and we use methods that account for the possibility that unmeasured factors exist that are correlated with education and drive both the likelihood of having the illness and the propensity to be diagnosed with illness if it exists. Data come from the National Health and Nutrition Examination Survey (NHANES) 1999-2008. Our findings show that education is not associated with diagnosed diabetes or hypertension, and it is positively associated with having been diagnosed with high cholesterol. However, when we consider both undiagnosed and diagnosed cases, there is a strong, negative association between education and having diabetes or hypertension. A small, positive association between education and high cholesterol persists, even when we include undiagnosed cases. When we account for the possibility of shared, unmeasured determinants of disease prevalence and disease diagnosis that are correlated with education, we find that for all three chronic conditions, education is negatively associated with having undiagnosed chronic disease.