The connection between schooling and health is well documented. An important empirical issue that needs to be examined, however, is whether schooling's effects are due to individual health knowledge differences. This empirical study examines this issue with an increasingly important health indicator, obesity. Since provision of health knowledge is a major tool of public agencies promoting health, this empirical study uses a new direct measure of health knowledge to test this hypothesis. The results show that knowledge is inversely related to the probability that an individual is obese. Schooling's effects on relative weight and the probability of being obese are explained by differences in knowledge. This result may imply that schooling's effect on the allocative efficiency of the household production of health is the main reason schooling is linked to health behaviour. The result also may imply that the most effective method of health education is to highlight the disease element of poor dietary habits and health. More importantly, the simulations conducted suggest positive returns to knowledge based on improvements in the probability estimates.