Educational attainment has been negatively associated with the prevalence, morbidity and mortality of many diseases. With knee osteoarthritis as an example, we used NHANES I data to examine whether the cross-sectional association between formal education and disease is due to known risk factors, and also whether educational attainment is more strongly associated with self-reported symptoms or with radiographic change. We found univariate associations between osteoarthritis and low levels of education. For radiographic knee osteoarthritis in women, and in both sexes combined, this relationship was explained by controlling for known risk factors, which included age, knee injury, race, obesity, and occupation. However, even after adjusting for these major risk factors and the presence of radiographic changes, reporting of knee pain and arthritis at any site remained significantly associated with low educational attainment, especially for those with less than or equal to 8 years of education.