High plasma levels of oestrogens are associated with increased breast cancer risk. If smoking, as has been suggested, have both a tumour initiating mutagenic effect and a protective anti-oestrogenic effect, one would assume that smokers who give up smoking have the highest incidence of breast cancer. This was evaluated in the follow-up of a cohort of 10,902 women of whom 4,359 were premenopausal. Record-linkage with official cancer registries yielded 416 incident cases during an average follow-up of 13.6 years. The adjusted relative risk in all ex-smokers was 1.31 (1.02-1.69), as compared to never smokers, and in premenopausal ex-smokers it was 1.57 (1.07-2.30). Breast cancer incidence in premenopausal ex-smokers was inversely related to time since cessation, (p for trend = 0.01), and was highest among the women who had given-up smoking less than 12 months before screening: 2.76 (1.55-4.91). There was no significant association between current smoking and breast cancer risk. We conclude that incidence of breast cancer in premenopausal women who have given up smoking is higher than it is in smokers and never smokers. To what extent this may be related to endocrine effects associated with smoking cessation remains to be evaluated.