The extent to which current, ex- and passive smoking are associated with other risk factors, and the potential for confounding arising from these associations, was studied using a representative sample of 9003 British adults. The distribution of 33 lifestyle factors generally considered associated with adverse health were compared in current smokers, ex-smokers, never smokers living with a smoker ("passive smokers") and other never smokers. Of the 33 risk factors 27 showed a significantly higher prevalence in heavy smokers than in never smokers and only two showed a lower prevalence. For many risk factors, prevalence increased with amount smoked, decreased with time of smoking cessation and was increased in passive smokers. The possible magnitude of bias from confounding by the risk factors is estimated. It is concluded that confounding by multiple risk factors may be an important issue in smoking studies where weak associations are observed. This applies particularly to studies investigating the possible association of passive smoking with various health effects.