High-resolution peripheral quantitative computed tomography (HR-pQCT) is increasingly being used in the research setting to assess the effects of osteoporosis treatments and disease on trabecular and cortical bone compartments. Further in-depth study of HR-pQCT measurement variables is essential to ensure study strength and statistical confidence when designing large multicenter studies. Duplicate HR-pQCT examinations of the distal radius and tibia were performed in 180 healthy men and women ages 16-18, 30-32, and >70 years. HR-pQCT images were processed using standard and extended cortical bone analysis techniques. Biomechanical properties of bone were assessed using finite element analysis. Percent root mean square coefficient of variation (RMSCV) was calculated for each measurement variable. Age, site, and gender influences on measurement variability were investigated using variance ratio tests. Smaller precision errors were observed for densitometric (0.2-5.5%) than for microstructural (1.2-7.0%), extended cortical bone (3.4-20.3%), and biomechanical (0.3-9.9%) measures at both the radius and tibia. Tibial measurements (RMSCVs = 0.2-7.4%) tended to be more precise than radial measurements (RMSCVs = 0.7-20.3%). Variability was influenced by age, site, and gender (all p < 0.05). HR-pQCT measurements for the tibia were more precise than those for the radius, and this may be explained by the larger bone volumes examined and the reduced likelihood of movement artifact. The greater measurement variability observed for older volunteers may be due to the loss of bone density and microstructural integrity with age.