The purpose of this study was to describe application and machine accuracy for a new computed tomography (CT) guided, frameless, stereotactic brain biopsy system in dogs. Heads from ten canine cadavers were secured to a bite-plate with six attached fiducial markers and imaged using CT. Fiducialized CT images were imported into stereotactic software and spherical phantom lesions between 3.9 and 5.5 mm in diameter were created in six locations. Infrared cameras and reflective markers were used to register fiducials to the reconstructed image set. Coordinates in theX,Y, andZplanes were identified for each lesion center. Iohexol (1.5 μl of 240 mgI/ml) was injected into the center of each lesion and CT scans were repeated. Pre- and postinjection CT images for each cadaver were fused using the system software. Application accuracy was calculated using the center of each phantom lesion and the center of each injected contrast material location. Machine accuracy was calculated using a phantom with known distances between four fixed points in theX,Y, andZplanes. Mean application accuracy in the first 5 cadavers was 4.3 mm (95% confidence interval [CI] 2.9–4.3 mm) and in the second 5 cadavers was 2.9 mm (95% CI 2–3.9 mm). The more superficial lesions were targeted significantly less accurately than the deeper lesions (P= 0.0183). Median machine accuracy was 0.1 mm and the range was 0.1–0.2 mm. Findings supported use of the new biopsy system for canine brain lesions >3.9 mm in diameter.