BACKGROUND External root resorption usually does not present a clinical sign or symptom, and, therefore, diagnosis is mainly based on radiographic examination. Many studies confirmed the advantage and accuracy of cone-beam computed tomography (CBCT) in evaluating root resorptions. We aimed to evaluate the diagnostic accuracy of CBCT images of chemically induced external root resorptions on extracted human teeth taken in different voxel sizes. MATERIAL AND METHODS In this in vitro study, 36 maxillary and 36 mandibular human incisor teeth, extracted owing to periodontal disease, were used. External resorption cavities were created on the buccal and proximal surfaces by using 10% hydrochloric acid with different application periods of 10, 30, and 60 min. Resorption cavities in different depths were induced to simulate different levels of external resorption. CBCT images were taken with Planmeca Promax 3D Max CBCT (Planmeca, Helsinki, Finland) in 4 different voxel sizes: 400, 200, 150, and 100 μm. RESULTS There was no statistically significant difference between interobserver and intraobserver reliability. Higher observer agreement was obtained for 100-μm and 150-μm voxel sizes. For detection of external root resorption defects, interobserver agreement was highest for the 100-μm voxel size and when defects were located on the proximal side of the samples. The highest k values were obtained for samples kept in hydrochloric acid for 60 min. CONCLUSIONS Chemically induced resorption cavities should be used for experimental studies to better imitate clinical conditions. CBCT requirement is still ambiguous for detection of external resorptions, and more experimental and clinical studies are needed.