Objective: To evaluate the reproducibility of four parameters for quantitatively assessing maxillary molar furcation involvement (FI) by cone beam computed tomography (CBCT). Methods: Thirty-nine sites with degree II FI, classified by probing of 21 maxillary molars, were investigated. Degrees of FI in these sites were assessed based on CBCT data. In these samples, four parameters for quantitatively assessing FI in CBCT images were measured. The parameters included horizontal bone loss at furcation entrance level (HBL), maximum HBL (HBL-max), maximum vertical bone loss (VBL-max) and root trunk length (RT). The reproducibility of the measurements was evaluated. Results: Amongst the 39 degree II FI classified by probing, only 17.9% were confirmed by CBCT. The other 46.2% were 'through and through' defects, 15.4% were fused roots and 20.5% were degree I FI in the CBCT image. The intraobserver repeatability for all four parameters was high, with intraclass correlation coefficients (ICC) of 0.960 for HBL, 0.992 for HBL-max, 0.987 for VBL-max and 0.983 for RT. The ICCs for two observers was also high (ICCs: 0.873 to 0.947). The parameters and related methods of measurements proposed in the study showed high reproducibility. CBCT images provided more details in assessing maxillary molar FI. Conclusion: The parameters and related methods of measurements developed in this study showed high reproducibility. CBCT images provide more details in assessing maxillary molar FI. Palabras clave: cone-beam computed tomography, furcation involvement, linear measurement, maxillary molars