Objective To study the change of Bruch's membrane opening-minimum rim width (BMO-MRW) inpatients with early diabetic retinopathy (DR). Methods This was an observational cross-sectional study. A total of 123 patients (208 eyes) with type 2 diabetes mellitus and 66 (121 eyes) healthy volunteers (normal control group) in our hospital from Mar. to Dec. 2020 were enrolled. According to the international clinical DR severity grading scale, type 2 diabetes mellitus patients were divided into non-DR group (n=72) and mild-DR group (n=51). Spectral-domain optical coherence tomography (SD-OCT) was used to perform 24 radial B-scan centered on the optic disc and peripapillary circular scan. All scanning results were obtained relative to the specific axis of the fovea-BMO center (FoBMO axis) of the eye. The values of BMO-MRW and circular retinal nerve fiber layer (RNFL) thickness on 24 B-scan lines were measured, and the data were analyzed in 6 sectors (superonasal, nasal, inferonasal, superotemporal, temporal, and inferotemporal) for statistical analysis. Results The mean BMO-MRW values and each sector BMO-MRW values of the normal control group, non-DR group and mild-DR group showed a decreasing trend. The mean BMO-MRW values of the non-DR group and mild-DR group were (304.64±36.67) μm and (299.39±41.27) μm, which were significantly thinner than that of the normal control group ([315.14±41.60] μm) (P=0.040 and 0.005, respectively). The BMO-MRW values in the superotemporal sector of the non-DR group and mild-DR group were (308.35±52.40) μm and (304.60±53.33) μm, which were significantly thinner than that of the normal control group ([324.82±52.40] μm) (P=0.012 and 0.005, respectively). The BMO-MRW values in the inferotemporal sector of the non-DR group and mild-DR group were (339.49±51.39) μm and (331.48±47.21) μm, which were also significantly thinner than that of the normal control group ([358.58±48.94] μm) (P=0.003 and P<0.001, respectively). The mean BMO-MRW was significantly positively correlated with the mean RNFL thickness (r=0.187, P<0.001). The correlation between BMO-MRW and RNFL thickness in each sector ranged from medium correlation (with the highest correlation in inferotemporal sector; r=0.333, P<0.001) to no correlation (in temporal sector; r=0.087, P=0.115). Conclusion The BMO-MRW of diabetic patients is thinner than that of normal controls, and it is positively correlated with the changes of RNFL thickness, suggesting that BMO-MRW can be used as an indicator of early neurodegenerative changes in DR.