We used 10 years of longitudinal data from Penn State Young Women's Health Study to explore predictors of adult bone structural geometry and strength. One hundred twelve participants were enrolled in the study at age 12. We report findings on the 76 participants who remained in the study for 10 years. Measurements were recorded biannually for the first 4 years and annually thereafter. Proximal femur DXA scans (Hologic QDR 2000) were taken from 17-22 years and analyzed using a hip structure analysis program to assess areal bone mineral density (BMD, g/cm2), subperiosteal width, cortical thickness, bone cross-sectional area (CSA), and section modulus (Z) at the narrow neck and femoral shaft. Total body lean mass (g) was measured with DXA total body scans. Nutrition, anthropometry, and sex steroids [testosterone (T) and estradiol (E2)] were measured from ages 12-22 years. Multiple regression models were used to assess predictors of change in bone variables (17-22 years) and absolute bone values (average of age 21 and 22 years, n = 79). Neck Z (+3.1%) and width (+1.3%), but not BMD (-0.8%), increased significantly from age 17 to 22 years. At the shaft, all variables increased (+1.0-4.0%, P < 0.01). After controlling for baseline (age 17) height, weight and bone measurement, weight change (neck) or lean mass (shaft), and age of menarche were the primary predictors of change in bone strength. After controlling for height and weight, only lean mass predicted absolute young adult Z at both the neck (r2 = 0.48, P < 0.01) and the shaft (r2 = 0.67, P < 0.01). When lean mass was removed from the model, sports exercise score replaced lean mass as a predictor of Z at both neck (r2 = 0.40, P < 0.01) and shaft (r2 = 0.60, P < 0.01) sites. For neck and shaft cortical thickness and BMD, both estradiol and sports score/lean mass were positive predictors (r2 = 0.15-0.40, P < 0.01). For neck bone width, testosterone levels (negative) and lean mass (positive) were significant (r2 = 0.48). Results were similar for each geometric variable at the shaft site. These data suggest that bone adapts its bending strength primarily to mechanical loading (represented by lean mass and sports exercise score) and that sex steroids are associated with bone geometric structure.