Recent studies from Finland reported that maternal pelvic size predicted daughters' breast and ovarian cancer, possibly because maternal pelvic size is a marker for in utero hormone exposure. We sought to replicate this association in 3,845 women born between 1915 and 1929 in Uppsala, Sweden, and followed from 1960 to 2002. Archived obstetric records provided the standard measures of maternal pelvic size (intercristal distance, interspinous distance, the intercristal-interspinous difference, and the external conjugate distance). The Swedish Cancer Registry ascertained cancer incidence, with 273 cohort members developing primary breast cancer, and 52 developing primary ovarian cancer during the follow-up period. There was no evidence (P > 0.1) of an association between any measure of maternal pelvic size and incidence of either breast or ovarian cancer. This was true both before and after adjustment for various characteristics of the women and their mothers, and in analyses stratified by age at diagnosis (<50 versus >or=50 years of age, as a proxy for premenopausal and postmenopausal ages). There was also no evidence of an association in subgroup analyses restricted specifically to those groups in which the Finnish data found the greatest effect. Our study is of comparable size to the Finnish studies and was highly powered (>99%) to detect effects of the magnitude they reported. Our nonreplication therefore casts doubt on the link between maternal pelvic size and risk of breast and ovarian cancer in the offspring.