Objectives We investigated the relationship between loading history and bone biomechanical properties used in physical activity reconstructions. These bone properties included bone bending and torsional strength (J), cortical area (CA), the direction of the major axis (theta angle), and element shape ratios determined from cross sections of standardized bone length. In addition, we explored the applicability of anatomically determined cross sections. Methods Our material consisted of hip and proximal thigh magnetic resonance images of Finnish female athletes (N = 91) engaged in high‐jump, triple‐jump, endurance running, swimming, power‐lifting, soccer and squash; along with a group of active non‐athlete individuals (N = 20). We used regression analysis for size‐adjustment, and the extracted residuals were then used to compare differences in the bone properties between groups. Results We found that triple‐jumpers, soccer players, and squash players had the greatest values in CA and J, swimmers and non‐athletes had the smallest, whereas high‐jumpers, power‐lifters, and endurance runners exhibited interim values. No between‐the‐group differences in element shape ratios or theta angles were found. We found that influences of activity were similar regardless of whether standardized length or anatomically determined cross sections were used. Conclusions Extreme (triple‐jump) and directionally inconsistent loading (soccer and squash) necessitate a more robust skeleton compared to directionally consistent loading (high‐jump, power‐lifting, and endurance running) or non‐impact loading (swimming and non‐athletes). However, not all of these relationships were statistically significant. Thus, information gained about physical activity using bone properties is informative but limited. Accounting for the limitations, the method is applicable on fragmented skeletal material as anatomically determined cross sections can be used.