In a nested case-control study (2001-2004), the authors investigated the association between mammographic density and breast cancer risk among women of Caucasian, Japanese, and Native Hawaiian ancestry in the Hawaii component of the Multiethnic Cohort Study. The authors retrieved several prediagnostic mammograms for breast cancer cases and for controls frequency-matched to cases by age and ethnicity. A reader who was blinded to case status and year of mammogram performed computer-assisted density assessment. Suitable mammographic readings were obtained for 607 cases and 667 controls. The authors used unconditional logistic regression to estimate odds ratios and 95% confidence intervals while adjusting for confounders. Mean percent density and mean dense area were significantly greater for cases than for controls: 39.6% vs. 29.7% and 37.3 cm2 vs. 28.4 cm2, respectively. For the earliest mammogram taken, the overall odds ratio for a 10% increase in breast density was 1.22 (95% confidence interval: 1.14, 1.30), and the overall odds ratio for a 10-cm2 increase in dense area was 1.17 (95% confidence interval: 1.11, 1.24). The similar sizes of the areas under the receiver operating characteristic curve (0.66) confirmed that percent density and dense area predicted breast cancer equally well. Because the risk estimates appeared higher for Caucasians and Native Hawaiians than for Japanese women, ethnicity-specific models may be necessary to predict risk from breast density in different ethnic groups.