Rates of smoking have decreased dramatically in most Northern European countries over the last fifty years or so, but this decline has not been uniform across the population and there have actually been increases in smoking among lower income and social class groups. Although smoking differentials cannot account for the wide social class inequalities in mortality and morbidity in these countries, they are a contributing factor. This paper argues that the social structuring of smoking rates suggests that social and economic processes may have a major role in starting and quitting behaviour. We test four hypotheses: The first holds that social class differentials in smoking reflect the direct impact of different levels of knowledge about the risks of smoking across educational groups. The second that social class differences reflect the indirect affect of educational differentials acting through educations influence on risk perception and future orientation. The third hypothesis also invokes future orientation, but attributes differences in this variable to socio-economic disadvantage. The last hypothesis holds that differential rates of smoking across social classes actually reflects the indirect affect of social deprivation on the push factors to smoke such as lack of control and psychological stress. Our analyses shows little support for the first hypothesis with knowledge differences accounting for no more than 10% of the class differential. Tests of the role of future orientation show that this plays almost no role. The last hypothesis gains most support. Measures of disadvantage and deprivation account for half of the differential in class smoking.