BACKGROUND Long bone structural parameters such as cross-sectional area or area moments of inertia are useful measures of long bone mechanical properties. We implemented a three-scan densitometric method to measure structural parameters in long bones of the lower leg in vivo. The validated method was applied to investigate the relationship between activity level, age, and long bone structural parameters in women. METHODS An aluminum phantom was used to estimate in vivo setup accuracy. In vivo precision was determined by same-day repeated measures on human subjects. For the activity study, women were recruited in two age groups (25-35, 60+yr) and two activity levels (recreational runners, nonrunners). Scans were taken of the middle third of the lower right leg; structural parameters for the tibia and fibula were determined at each scan line, averaged over the section, and adjusted by factors accounting for body size variations. RESULTS Aluminum phantom cross-sectional area was underestimated by 4-6%, principal moments were underestimated by <5%, and principal angles were within +/-1.2 degrees. In vivo precision results (lower energy, scans spanning 60 degrees) indicated coefficients of variation for cross-sectional area (A), principal moments of inertia (Imax, Imin), and polar moment of inertia (J) of 0.52, 5.87, 2.22, and 3.82%, respectively. The activity study showed mean adjusted tibial A, Imax, Imin, and J were significantly higher in runners compared with nonrunners. There was no dependence on age. CONCLUSIONS A three-scan densitometric method for measuring cross-sectional structural parameters in long bones in vivo was validated; accuracy and precision measurements establishes confidence limits. From the activity study results, we postulate that higher loads associated with running lead to increased cross-sectional parameters to support axial loads, bending, and torsion in the tibia.