Abstract Introduction Mammographic density is a strong risk factor for breast cancer. Our objective was to examine its association with polymorphisms identifying breast cancer susceptibility loci that were ascertained in recent genome-wide association studies. Methods Subjects were 825 women who participated in previous case–control studies of mammographic density and genetic factors nested within the Multiethnic Cohort study and were from three ethnic groups (White, Japanese American, Native Hawaiian). Eight polymorphisms (rs2981582 in FGFR2, rs3803662 and rs12443621in TOX3, rs3817198 in LSP1, rs981782 and rs10941679 near HCN1/MRPS30, rs889312 in MAP3K1, and rs13387042 at 2q) were examined. Mammographic density was quantified with a computer-assisted method as the percent dense area: the area of radiologically dense fibroglandular tissue relative to the total breast area that also includes radiologically lucent fatty tissue. Results The polymorphism rs12443621 in TOX3 was associated with percent dense area; women with at least one G allele (previously associated with increased breast cancer risk) had 3% to 4% higher densities than women with two A alleles. The polymorphism rs10941679 near HCN1/MRPS30 was also associated with percent dense area; women who were homozygous for the G allele (previously associated with increased breast cancer risk) had 4% to 5% lower densities than women with at least one A allele. The other polymorphisms were not associated with percent dense area. Conclusions The available data suggest that the effects of most of these polymorphisms on breast cancer are not mediated by mammographic density. Some effects may have been too small to be detected. The association with rs12443621 may provide clues as to how variation in TOX3 influences breast cancer risk.