Striking geographic variation in death rates for coronary heart disease has been observed to occur within the United States. Coronary disease death rates for states are strongly correlated with per capita cigarette sales, estimated from state cigarette tax revenues. The correlation noted was as high as +0.76 if only the states with the most accurate cigarette consumption information were used. This correlation could not be attributed either to urbanization or to softness of the local water supply. It is suggested that accompanying the search for new hypotheses to explain geographic variation in coronary heart disease mortality should be an attempt to determine whether any of the established risk factors are responsible.