OBJECTIVES: To investigate the root canal morphology of mesiobuccal (MB) roots in maxillary first molars, and to assess the presence of a second mesiobuccal canal (MB2) according to the location of the main MB canal. MATERIALS AND METHODS: A total of 72 extracted permanent maxillary first molars were collected from dental clinics and were scanned with micro-CT and reconstructed three-dimensionally. The root canal systems were recorded according to Vertucci's classification, and the occurrence of accessory canals was also recorded. The root canal dimensions were measured at the coronal (furcation plane), middle, and apical root levels. The long (D) and short (d) diameters as well as the palatal (P) and buccal (B) distances from the center of the first mesiobuccal canal (MB1) to the root surface were measured, and the ratios of D/d and P/B were calculated. Receiver operating characteristic (ROC) curve analysis was used to evaluate the diagnostic accuracy of using the ratio of P/B for predicting the presence of an MB2 canal. The best cut-off point was determined according to the sensitivity and specificity. RESULTS: The MB roots most frequently had a type 2-2 root canal with an incidence of 37.5% (27/72), followed by the type 1-1 (23.6%, 17/72) and type 2-1 (16.7%, 12/72) canal forms. Type 1-2 canals were detected only in 5 molars (6.9%), and type 2-1-2 canals were detected in 6 molars (8.3%). The other 5 cases included 1 case of type 1-2-1 canal and 4 cases of triple canals. MB2 canals were detected in 76.4% (55/72) of the total sample teeth. The incidence of accessory canals was 56.9% (41/72). The mean ratio of D/d was generally "greatest to least": coronal level > middle level > apical level for different root levels and MB single > MB1 > MB2 for different canals, which reflected a trend from a flat to a circular cross-sectional shape. ROC curve analysis showed that at the coronal and middle root levels, areas under the ROC curve (AUC) were greater than 0.99 (P < 0.01), and the best cut-off point was 1.58 and 1.55, respectively; at the apical level, the AUC was 0.94 (P < 0.01), and the best cut-off point was 1.77. CONCLUSIONS: The MB2 canals may be present in the MB roots of maxillary first molars with a high occurrence rate at various levels, and the P/B ratio of the MB1 is a good index for predicting the presence of an MB2. However, since all the sample teeth were collected from a Chinese population, clinicians have to be cautious while trying to apply the conclusions on teeth of other ethnic populations. CLINICAL RELEVANCE: By calculating the P/B ratio, an index reflecting the buccal deviation of the MB1, clinicians can predict the presence of an invisible MB2 in cone-beam computed tomography images with inadequate resolution.