Objective: The aim of this study was to investigate the accuracy of endoscopic visualization to detect root canal anastomoses at the coronal half of the mesial root canals of mandibular molars using micro-computed tomographic (micro-CT) images as reference. Material and methods: Seventy-four mesial roots of mandibular first molars with (n = 47) or without (n = 27) intercanal anastomosis were selected based on the micro-CT scans of 269 mandibular first molars at a pixel size of 10 mu m. The specimens were mounted on the mannequins and their root canals were evaluated using dental operating microscope (DOM) and endoscope. The endoscopic probe was inserted into the main mesial root canals and 2 blinded observers evaluated the presence of a divergence point of anastomosis (where the branching occurs) as 'present' or 'absent'. The scorings were compared with the three-dimensional reconstructed images of the specimens and recorded as 'correct' or 'incorrect' evaluation. Degree of agreement between evaluators was assessed with Kappa test and the accuracy of endoscopic visualization according to the presence and location of anastomosis was compared using Fisher exact tests with a significance threshold at 5%. Results: High inter-examiner reliability was reported (0.91). None of the divergence points were identified using DOM whereas 11 divergence points were detected using endoscope, corresponding the 23.4% of the intercanal anastomoses. The endoscope also showed the absence of an intercanal anastomosis correctly in all of the specimens without an anastomosis. Detectability of a divergence point using endoscope was not affected by its location within the coronal half of root canal (p > .05). Conclusions: The endoscopes were able to visualize the divergence points of 23.4% of the intercanal anastomoses located at the coronal halves of root canals.