Age changes in bone strength during growth have important consequences for predicting fracture risk and identifying metabolic bone diseases in children. Previous investigations of bone strength changes in juveniles have been limited to a cross-sectional study design. The present study utilizes longitudinal data collected over the entire growth period to examine age changes in femoral and humeral strength in a sample of 20 individuals (10 males, 10 females). The sample was obtained from the archives of the Denver Child Research Study, a longitudinal study carried out in the 1940's-1960's. The original study included radiography of the limbs performed at 6 month-1 year intervals from near birth through late adolescence (average 34.5 measurements/individual). Periosteal and cortical bone breadths were taken at femoral midshaft and at 40% of bone length from the distal end of the humerus, and used to calculate section moduli, measures of bending and torsional strength. Soft tissue breadths, also derived radiographically, were used to calculate muscle areas in the upper and lower limbs. Body weights and statures at each examination were available from the study archives. A mechanically appropriate "body size" parameter was calculated as body weight x bone length (femoral or humeral). Raw data were smoothed using LOWESS, and growth velocities calculated between each age interval. Growth velocities for femoral strength are strongly correlated with those for body size (r^2 = .65-.80), but very poorly correlated with those for stature (r^2<.06). Growth velocities for humeral strength are moderately correlated with body size (r^2=.40-.73), and again very poorly correlated with stature (r^2<.05). Bone strength and muscle area velocities are weakly but significantly correlated (r^2=.10-.25), except in the male upper limb, where the correlation is much stronger (r^2=.65). Age at adolescent peak growth velocity for stature is significantly earlier than those for body size, muscle areas, or bone strengths, which are not significantly different from each other. Thus, relative to a mechanically appropriate measure of body size (body weight x bone length), there is no apparent "lag" in bone strength during early adolescence. If bone strength is expressed relative to body size, only a minority of the sample shows evidence of a minimum reached in early or mid-adolescence. These results are different from those obtained in a study of the distal radius, and suggest region-specific variability in growth, perhaps mechanically mediated. Overall, the results here argue strongly for the importance of mechanical factors -body weight and (especially in the male upper limb) muscular loadings -in the development of bone strength prior to adulthood.