The causation of breast cancer in certain strains of mice by a virus that can be transmitted vertically, through the milk produced during lactation, has led to the hypothesis that a similar phenomenon could exist in humans. There have been laboratory-based studies in humans suggesting that a virus may be involved in the etiology of female breast cancer although other investigations did not support this hypothesis. Descriptive data and epidemiologic evidence of ecologic nature do not indicate a role of lactation in the causation of human breast cancer, but the hypothesis has not been adequately assessed in analytic epidemiologic studies. A nested case-control study undertaken in Sweden to examine the role of prenatal factors on breast cancer risk in the offspring, allowed the evaluation of the importance of breast-feeding in the causation of this disease. Standardised records concerning women born at the Uppsala University Hospital from 1874 to 1954 were linked with invasive breast cancer incident cases, identified through their unique national registration number in the Swedish Cancer Registry during 1958-1990. For each case with breast cancer, the females born to the first three mothers admitted after the case's mother were selected as potential matching controls. Only controls living in Sweden and free from breast cancer until the time of diagnosis of breast cancer in the corresponding case were eventually included in the study. The analysis was based on 458 cases of breast cancer born in singleton pregnancies and 1,197 singleton age- and birth date-matched controls. Breast-feeding was not a significant or suggestive risk factor for breast cancer in the offspring; compared to women who at discharge were wholly or partly breastfed, women who as newborn were not breastfed had a relative risk of breast cancer of 0.97 with 95% confidence interval 0.44-2.17 (P = 0.95).