This study aims to evaluate the ability of C-arm cone-beam CT to detect intracranial hematomas in canine models. Twenty one healthy canines were divided into seven groups and each group had three animals. Autologous blood and contrast agent (3 mL) were slowly injected into the left/right frontal lobes of each animal. Canines in the first group, the control group, were only injected with autologous blood without contrast agent. Each animal in all the 7 groups was scanned with C-arm cone-beam CT and multislice computed tomography (MSCT) after 5 minutes. The attenuation values and their standard deviations of the hematoma and uniformed brain tissues were measured to calculate the image noise, signal to noise ratio (SNR) and contrast to noise ratio (CNR). A scale with scores 1-3 was used to rate the quality of the reconstructed image of different hematoma as a subjective evaluation, and all the experimental data were processed with statistical treatment. The results revealed that when the density of hematoma was less than 65 HU, hematomata were not very clear on C-arm CT images, and when the density of hematoma was more than 65 HU, hematomata showed clearly on both C-arm CT and MSCT images and the scores of them were close. The coherence between the two physicians was very reliable. The same results were obtained with C-arm cone-beam CT and MSCT grades in measuring SD value, SNR, and CNR. The reasonable choice of density detection range of intracranial hematoma with C-arm cone-beam CT could be effectively applied to monitoring the intracranial hemorrhage during interventional diagnosis and treatment.