Abstract: To accurately characterize the complex pore structure of shale and address the uncertainty of conversion coefficients in shale reservoir pore size characterization using nuclear magnetic resonance (NMR) technology due to lithofacies differences, typical matrix-type continental shale samples from Paleogene Shahejie Formation in Dongying Sag of Bohai Bay Basin were selected. Based on observations of cores, thin sections, and scanning electron microscopy analysis, the NMR T₂ spectrum distribution characteristics of shale in different lithofacies in original, washed-oil, saturated-oil, and base-removed states were systematically compared. The pore size of NMR T₂ spectrum was calibrated using full pore size distribution data obtained from combined high-pressure mercury injection-gas adsorption characterization, clarifying the range of pore size conversion coefficients for different lithofacies and multi-scale pore structure characteristics. The NMR radius conversion coefficient of continental matrix-type shale was not a fixed value, ranging from 4 to 35 nm/ms. The conversion coefficient was closely related to shale mineral composition, structural characteristics, and sedimentary structures. Generally, the more developed the laminae and the more complex the structure, the higher the coefficient value. Shale reservoir pores were mainly mesopores and macropores with pore size of 2-500 nm. Among them, laminated shale exhibited the best storage performance due to more developed macropores and microfractures. The differences in pore structure and reservoir quality were controlled by the synergistic mechanism of "sedimentary structure and diagenetic crystallization". The more developed the laminae and the more complex the mineral structure, the better the pore structure, resulting in larger fluid storage space and more effective seepage channels. This study provides key parameters and methods for accurately characterizing the complex pore structure of shale using NMR technology, which is of great significance for the precise evaluation and "sweet spot" prediction of shale reservoirs.