INTRODUCTION: Vertical root fracture (VRF) in root-canal-treated teeth frequently results in tooth loss, partly because VRFs are difficult to diagnose and when detected the fracture is often beyond the point of preservation with surgical intervention. Nonionizing magnetic resonance imaging (MRI) has demonstrated the ability to detect small VRFs, but it is unknown how its diagnostic capabilities compare with the current imaging standard for VRF detection, cone-beam computed tomography (CBCT). This investigation aimed to compare the sensitivity and specificity between MRI and CBCT for detecting VRF, using micro-computed tomography (microCT) as a reference. METHODS: A total of 120 extracted human tooth roots were root canal treated using common techniques, and VRFs were mechanically induced in a proportion. Samples were imaged using MRI, CBCT, and microCT. Axial MRI and CBCT images were examined by 3 board-certified endodontists, who evaluated VRF status (yes/no) and gave a confidence assessment for that decision, from which a receiver operating characteristic curve was generated. Intra- and inter-rater reliability were calculated, sensitivity and specificity, and area under the curve. RESULTS: Intra-rater reliability was 0.29-0.48 for MRI and 0.30-0.44 for CBCT. Inter-rater reliability for MRI was 0.37 and for CBCT 0.49. Sensitivity was 0.66 (95% confidence interval [CI], 0.53-0.78) and 0.58 (95% CI, 0.45-0.70), and specificity 0.72 (95% CI, 0.58-0.83) and 0.87 (95% CI, 0.75-0.95) for MRI and CBCT, respectively. Area under the curve was 0.74 (95% CI, 0.65-0.83) for MRI and 0.75 (95% CI, 0.66-0.84) for CBCT. CONCLUSIONS: There was no significant difference in sensitivity or specificity between MRI and CBCT in detecting VRF, despite the early-stage development of MRI.