In a study in 29 health centre districts in Japan 91 540 non-smoking wives aged 40 and above were followed up for 14 years (1966-79), and standardised mortality rates for lung cancer were assessed according to the smoking habits of their husbands. Wives of heavy smokers were found to have a higher risk of developing lung cancer and a dose-response relation was observed. The relation between the husband's smoking and the wife's risk of developing lung cancer showed a similar pattern when analysed by age and occupation of the husband. The risk was particularly great in agricultural families when the husbands were aged 40-59 at enrolment. The husbands' smoking habit did not affect their wives' risk of dying from other disease such as stomach cancer, cervical cancer, and ischaemic heart disease. The risk of developing emphysema and asthma seemed to be higher in non-smoking wives of heavy smokers but the effect was not statistically significant. The husband's drinking habit seemed to have no effect on any causes of death in their wives, including lung cancer. These results indicate the possible importance of passive or indirect smoking as one of the causal factors of lung cancer. They also appear to explain the long-standing riddle of why many women develop lung cancer although they themselves are non-smokers. These results also cast doubt on the practice of assessing the relative risk of developing lung cancer in smokers by comparing them with non-smokers.