A strong pattern in smoking behavior can be demonstrated, in which smoking is much more prevalent among those occupational groups (and social strata) that are also more exposed to hazards in the workplace and much less prevalent among those groups less exposed to such hazards. As a consequence, comparing individuals with greater to those with lesser exposure to tobacco also compares groups that differ with respect to occupational exposure to dust, fumes and toxic substances and with respect to occupationally related lifestyle factors. Analyses of the U.S. National Health Interview Survey show that smoking and occupation are substantially confounded among individuals differing by (1) amount of smoking; (2) smoking cessation; (3) types of cigarettes smoked; (4) age of starting to smoke; and (5) exposure to more or less environmental tobacco smoke at home. This confounding between types of work and proximity to tobacco smoke may have masked relationships between type of employment and disease. But it is difficult to disentangle the effects of occupation and of smoking from each other without well planned further studies because (1) of the difficulty of estimating occupational effects and simultaneously adjusting for healthy worker effects, (2) satisfactory techniques for estimating relative effects of intertwined variables make demands on the quality and quantity of data that are not met by presently available data, and (3) there may be deeply rooted social and psychological attitudes toward effects of work versus effects of lifestyles that tend to bias investigative work.