We hypothesized that measures of physical activity would have a closer relationship with section modulus (SM), an indicator of bending resistance, than with bone mineral density (BMD) because physical activity might expand the bony envelope, which tends to reduce BMD for a constant bone mineral content. Four hundred twenty-three men and 436 women (mean age 72 years, SD =3) were recruited from a prospective population-based cohort study to a study of hip bone loss. Hip BMD was measured on two occasions 2-5 years apart (mean 2.7, DXA-Hologic 1,000 W). Hip structural analysis (HSA) software was used to calculate SM and BMD from the DXA scans on three narrow regions: the narrow neck (NN), intertrochanter (IT) and shaft (S). A physical activity and lifestyle questionnaire was administered at baseline. Multivariate repeated measures analysis of variance was used to model the associations between personal attributes (weight, height, age), physical activity and lifestyle variables with SM, cross-sectional area (CSA), sub-periosteal diameter (PD) and BMD. Men and women were analysed together after tests for interactions with gender, which were found not to be significant. In all regions female gender was associated with having lower values of all outcomes, and body weight was positively associated with all outcomes, i.e., SM, CSA, PD and BMD ( P<0.0001). Sub-periosteal diameter was positively associated with reported lifetime physical activity (IT and S, P<0.0001). There was a significant decline of BMD with age at the NN and S regions ( P<0.026), and the PD increased with age (NN and S, P<0.019). Previous fracture history was associated with having lower values of BMD, SM and CSA (except for S; P<0.022). Both section modulus and CSA were positively associated with heavy physical activity after age 50 years in all regions ( P<0.019), whereas NN BMD was the only BMD associate of heavy physical activity after 50 ( P=0.036). Time spent per week on recreational activities classified as no impact activity was positively associated with BMD, CSA and SM (multivariate P<0.016). In conclusion, proximal femur diameter is associated positively with reported life-long physical activity. If this is mediated through a loading related effect on sub-periosteal expansion, BMD would be an unsatisfactory outcome measure in physical activity studies since it is inversely related to projected bone area. SM in contrast was associated with several measures of recent physical activity and relates more directly to the bending experienced by the proximal femur in response to a given load. These data are consistent with an effect of mechanical loading to regulate bone strength through an anabolic effect maximal in the subperiosteal cortex, where the highest loading-related strains are experienced.