Bite force is a measure of feeding performance used to elucidate links between animal morphology, ecology, and fitness. Obtaining live individuals for in vivo bite-force measurements or freshly deceased specimens for bite force modeling is challenging for many species. Thomason's dry skull method for mammals relies solely on osteological specimens and, therefore, presents an advantageous approach that enables researchers to estimate and compare bite forces across extant and even extinct species. However, how accurately the dry skull method estimates physiological cross-sectional area (PCSA) of the jaw adductor muscles and theoretical bite force has rarely been tested. Here, we use an ontogenetic series of southern sea otters (Enhydra lutris nereis) to test the hypothesis that skeletomuscular traits estimated from the dry skull method accurately predicts test traits derived from dissection-based biomechanical modeling. Although variables from these two methods exhibited strong positive relationships across ontogeny, we found that the dry skull method overestimates PCSA of the masseter and underestimates PCSA of the temporalis. Jaw adductor in-levers for both jaw muscles and overall bite force are overestimated. Surprisingly, we reveal that sexual dimorphism in craniomandibular shape affects temporalis PCSA estimations; the dry skull method predicted female temporalis PCSA well but underestimates male temporalis PCSA across ontogeny. These results highlight the importance of accounting for sexual dimorphism and other intraspecific variation when using the dry skull method. Together, we found the dry skull method provides an underestimation of bite force over ontogeny and that the underlying anatomical components driving bite force may be misrepresented.