The present study aims to determine if morphological differences of the scaphoid, trapezoid, and second metacarpal are associated with thumb osteoarthritis by comparing three-dimensional bone models of healthy controls and osteoarthritis patients. Fifty-eight patients with moderate to severe thumb osteoarthritis (modified Eaton ≥ II) and 35 healthy controls from a larger completed investigation were examined. To quantify morphological variations, a statistical shape model was created that classified binary with respect to the Eaton grading system:​ non-osteoarthritis or moderate/severe osteoarthritis (II-IV). Three-dimensional surface models based on computed tomography images from the scaphoid, trapezoid, and second metacarpal were used to “train” the statistical shape model and yielded features that best explain variation within the three bones:​ the principal components These principal components were tested for significant differences between patient and control group. Additionally, a statistical shape model entailing all three bones was created. For the second metacarpal, only a single principal component was significantly associated with osteoarthritis (p = 0.035). The combined model utilizing all three bones, however, showed that with using one principal component of each of the bones, we could classify a sample as moderate/severe trapeziometacarpal osteoarthritis with an accuracy of 0.77. No individual shape components of the scaphoid or trapezoid significantly correlated to osteoarthritis. This study affirms that basilar thumb osteoarthritis is not limited to the trapeziometacarpal joint. Future studies investigating surrounding bones and joints as contributors to disease occurrence or progression will provide a more holistic insight into the prevention, diagnostic, and treatment of thumb osteoarthritis.