Recent evidence suggests that estrogen replacement therapy (ERT) may be effective in preventing bone loss even in late postmenopausal women. We analyzed the role of age and other factors as determinants of ERT action on vertebral bone density (VBD) in 73 consecutive Caucasian postmenopausal women treated for 1 yr or longer. A group of 44 women who received calcium supplements only were analyzed as controls. VBD was measured every 6 months by dual energy x-ray absorptiometry, and yearly rates of change were calculated by linear regression for each subject. VBD increased in the estrogen-treated group (+17.32 ± 2.84 mg/cm2.yr; 2.42 ± 0.37%/yr) and did not significantly change in untreated subjects (-4.08 ± 3.68 mg/cm2/yr; -0.60 ± 0.58%/yr). Women older than 60 yr experienced greater, although not significant, increments compared to younger subjects (+3.23 ± 4.03% vs. +1.42 ± 3.00%/yr), as did women more than 10 yr postmenopausal compared to patients within 10 yr after the menopause (+3.38 ± 4.11% vs. +1.19 ± 2.70%/yr). Rates of VBD change were positively correlated with age (r = 0.29; P = 0.014), years since menopause (r = 0.33; P = 0.005), and body mass index (BMI; r = 0.35; P = 0.003) and negatively with estimated initial VBD (r = -0.23; P = 0.05). However, only the relationships between bone density changes and BMI (r = 0.33; P < 0.01) and estimated initial VBD (r = -0.26; P < 0.05) held in a partial correlation analysis. BMI and estimated initial VBD were the only significant predictive factors for response to ERT (r2 = 21%) in a multivariate regression model. Therefore, the response to ERT depends mostly on BMI and initial VBD. Women with large body frames and those with lower initial VBD respond better to estrogen than slender women with higher bone mass.