The purpose of this study was to develop an optimized imaging protocol for canine dental imaging using single-detector row CT and multidetector row CT. Two medium-sized mesaticephalic cadaver dog heads were scanned using varying slice thickness, sequential vs. helical scanning mode, sequential CT table increment, pitch, image reconstruction algorithm, and an additional moderate edge enhancement filter. For each series, two reviewers scored dental roots as not visible (score: 0), partially visible (score: 1) or completely visible (score: 2). Sharpness of the dento-alveolar margin was scored as blurry (score: 0), sharp (score: 1) or very sharp (score: 2). Consensus reader scores were compared among CT protocols. For single-detector row CT, complete visualization of all tooth roots was only achieved with 1 mm slice thickness and image interval sequential or helical series (pitch of 1 or 2) using a high-frequency image reconstruction algorithm with or without additional edge enhancement. For four-detector row CT, complete visibility of all tooth roots was recorded in all series. For single-detector row CT, all medium frequency algorithm series yielded blurry margin scores and all other series yielded sharp margin scores. For four-detector row CT, sequential and helical 0.5 mm thickness images yielded very sharp dento-alveolar margin scores, whereas the 1 mm series yielded only sharp margin scores. Authors concluded that the optimal dental imaging protocol was a sequential mode 1 mm slice thickness and interval with a high-frequency image reconstruction algorithm and an additional moderate edge enhancement filter.