The UK National Case-Control Study Group has examined the relationship between smoking (both own smoking and passive), alcohol consumption and caffeine consumption and the risk of breast cancer. A total of 755 women with breast cancer diagnosed before the age of 36, each with an age-matched general population control, were interviewed, and detailed information on reproductive, contraceptive and medical history, personal attributes and habits were obtained. Additional data on passive smoking were obtained from a subgroup of women. There was no evidence of a statistically significant difference in breast cancer risk between subjects who had ever smoked as much as one cigarette per day and those who had not [relative risk (RR) = 1.01, 95% confidence interval (CI) 0.81-1.26]. Most relative risks for passive smoking exceeded unity, but there was little evidence of significant trends with increasing exposure. The lack of effect of own smoking, and the fact that such smokers are also themselves exposed to the effects of passive smoking, makes any relationship between exposure to others' smoking and breast cancer risk implausible. Alcohol consumption during the year prior to diagnosis and at ages 18 and 25 was examined. Consumers of 0.1-4.9 and 5.0-14.9 g per day generally had non-significantly increased risks compared with never drinkers, but consumers of more than 15 g per day had reduced risks.