Summary

Interpretational problems presented by an association between a habit (such a smoking) and a disease (such as lung cancer) are reviewed in broad terms. Mortality ratios ( " current cigarette smokers ' versus ' non- and occasional-smokers ') for ten cancer sites (Kahn, 1966) are considered in relation to age-standardized death-rates in England and Wales, over the period 1901-1954. Secular changes in sex-specific death-rates, for each site, do not agree with the hypothesis that they were caused predominantly by the increase in cigarette consumption. The sex- and age-dependence of disease is discussed with reference to lung cancer. Data for England and Wales, 1901-1970, are analysed by quinquennia. Their character-istics do not support a causal hypothesis of the association between cigarette smoking and lung cancer. Methods of distinguishing between causal and constitutional hypotheses of such associations are discussed. Special emphasis is placed on the value of large-scale studies of twin pairs discordant for the smoking habit