Detection and accurate classification of traumatic tarsal fractures are important for identifying cases requiring surgical intervention. The aim of this prospective, experimental, methods comparison study was to directly compare the accuracy, sensitivity, and specificity of tarsal computed tomography (CT), ten-view and two-view digital radiographs for detecting traumatic fractures of the canine tarsus. The working hypothesis was that tarsal fractures would be detected with higher accuracy, sensitivity, and specificity using CT imaging compared to radiography, and a tenview would be superior to a two-view radiographic study. Ten cadaver hind limbs of medium to large dogs received a CT scan and ten-view radiographic study before and after induction of fractures with a hydraulic press. All bones included in the radiographic images were assessed for fractures by two observers and gross dissection was used as the gold standard. The two-view radiographic study (dorsoplantar, lateromedial) was created from the ten-view study and reviewed 2 years later. All limbs sustained fractures, the most common locations were the talus and calcaneus (n = 7). The sensitivity of CT was greater than ten-view radiographic study (77% vs. 57%), while the specificity was similar (97% vs. 98%). The sensitivity and specificity of the ten-view and twoview radiograph studies were similar (57% vs. 55%; both 98%). Computed tomography images were reassessed postdissection to determine if failure to identify fractures resulted from observer error. Overall, CT was better than radiography for detecting fractures of the canine tarsus, however there was little improvement with ten-view compared to two-view radiographic studies.