This study uses measures of cross-sectional robusticity and asymmetry (based on humeral areal and inertial cross-sectional components) to test a prediction from bone remodeling theory that a physically active 18th century Quebec prisoner of war sample (N = 25) should have more robust and asymmetrical humeri than a nonphysically active 20th century New Mexico suburbanite sample (N = 27). Narrative accounts document that prisoners of war engaged in labor-intensive activities, and these activities were confirmed by observations of osteoarthritis and other pathologies. The suburbanite sample, for the most part, did not engage in such activities. The prisoners had higher levels of pathology than the suburbanites (e.g., 80% vs. 22% osteoarthritis; F = 17.95, P < 0.01). For robusticity, the populations did not differ significantly in total area, cortical area, moment areas of inertia about the mediolateral plane, or polar moment area of inertia. The Quebec prison sample did have significantly higher values for moment areas of inertia about the anteroposterior plane. For asymmetry, the populations did not differ in any values (total area, cortical area, moment areas of inertia about the mediolateral plane, moment areas of inertia about the anteroposterior plane, or polar moment of inertia). Thus, examinations of crosssectional robusticity and asymmetry failed to conclusively confirm the hypothesis that intensive labor leads to changes in humeral morphology. Possible explanations for the lack of differences are discussed, such as poor diet impeding bone remodeling. Nevertheless, the one significant finding suggests that cross-sectional shape is more useful in reconstructing activity patterns than amount of bone in a cross section. Results from this study join those from other recent investigations to suggest that additional controls are required before cross-sectional differences may be confidently attributed to activity patterns.