PURPOSE: Mammographic density has been strongly associated with breast cancer risk in several studies. To clarify possible pathogenic mechanisms underlying this relationship, we assessed factors affecting mammographic density in postmenopausal women.METHODS: Percent density (PD) was measured on pre-randomization mammograms from 425 participants in the Women's Health Initiative clinical trial; no women were currently using HRT. Trained observers used a previously validated computer-assisted thresholding technique to measure PD (ratio of dense areas to breast area) on craniocaudal films. Univariate relationships were assessed between PD and age, parity, education, history of HRT use, age at first birth, abortion history, alcohol use, serum cholesterol, physical activity, smoking, dietary fat intake, body-mass index (BMI), and waist-to-hip ratio (WHR). Multivariable regression was used to identify predictors of PD (log transformed), controlling for confounders.RESULTS: PD ranged from 0-60% with mean 8%. Age, gravidity, BMI, and WHR were negatively associated with PD. Mean PD in the highest tertiles of BMI and WHR was nearly half that in the lowest. Ethnicity (50% white, 28% black, and 19% Hispanic) and hysterectomy status (50% had intact uterus) were unassociated with PD. Controlling for age, BMI, WHR, smoking, hysterectomy status, prior HRT, and nulliparity, mean PD differed significantly by ethnicity, with black women having highest PD. However, the effect of prior HRT use differed by uterine status. In women with uteri intact, ethnicity, but not prior HRT use, was related to PD after controlling for age, BMI, and WHR. In hysterectomized women, ethnicity was unrelated to PD, while previous HRT was associated with lower PD.CONCLUSIONS: Gravidity, ethnicity and body shape and size are associated with mammographic density in postmenopausal women. Previous hormone use predicts lower PD, but only among hysterectomized women. This may be due to the higher frequency of oophorectomies in this group, which may lower PD and increase likelihood of HRT use. Planned analyses of entire sample of 1050 women will clarify the effect of these and other factors on density, and on density change in followup mammograms.