Birth weight is a significant predictor of breast cancer risk in adult life and mammary gland mass could be an intermediate stage in this long process. We have studied the association of birth size measurements with mammographic density, a marker of mammary gland mass. For a population-based sample of 893 postmenopausal women without previous cancer in Sweden, we retrieved information on birth size from birth records and their most recent mammography. Film mammograms of the medio-lateral oblique view were digitized and the Cumulus software was used for computer-assisted semi-automated thresholding of mammographic density. Results were analyzed using generalized linear models controlling for possible confounders. Mean percent mammographic density increased when comparing the extreme categories of birth weight (from 15.6% to 18.6%) and head circumference (from 15.5% to 20.4%), and the corresponding linear trends were statistically significant (p values 0.02 and 0.007, respectively). The associations were particularly strong when the cutoff for high versus low mammographic density was set at the relatively high value of 50%. Compared to women weighing 3001-3500 grams at birth, women with birth weights >4000g were at almost 3-fold risk of developing high mammographic density (odds ratio: 2.9, 95% confidence interval 1.1 to 7.9). No association with mammographic density was evident with respect to birth length which, however, is known to be less accurately measured. These results indicate that adult breast density, a powerful predictor of breast cancer risk, has intrauterine roots, as reflected in birth size.