Abstract: The Upper Triassic Xujiahe Formation in the Nanjiang area of the piedmont belt in the northeast Sichuan Basin hosts large-scale, typical conglomerate gas-bearing reservoirs. These reservoirs remain underexplored and hold significant potential. Among them, the fourth member of the Xujiahe Formation is an important gas-bearing reservoir characterized by complex structural and lithological features, low sensitivity of geophysical parameters, and strong multi-solution, making reservoir characterization highly challenging. The distribution characteristics and patterns of favorable reservoirs remain unclear. To address this, a prediction method tailored for early-stage conglomerate reservoir exploration was developed by integrating seismic forward modeling, seismic inversion, and sequential Gaussian simulation-based lithofacies modeling. The probabilistic body lithofacies modeling-based reservoir prediction method applied seismic forward modeling to determine reservoir boundaries, seismic inversion for preliminary spatial reservoir distribution prediction, as well as well-logging and inversion collaborative modeling for refined prediction of the spatial probabilistic distribution of lithofacies, ultimately achieving precise reservoir prediction. The results showed that this prediction method used lithofacies modeling as a bridge to combine logging data with seismic inversion results, enabling quantitative prediction of reservoir thickness. The combination with seismic forward modeling further refined reservoir boundaries, significantly enhancing prediction accuracy. Applying this method to the Nanjiang area, the distribution of conglomerate reservoirs in the fourth member of the Xujiahe Formation was finely characterized. The favorable reservoirs are mainly located in the braided channels of the conglomerate facies, characterized by medium to weak amplitudes and slight amplitude downcutting on seismic profiles. Reservoir quality improves with increasing sandy content and decreasing mud and conglomerate content, resulting in weaker seismic amplitudes. The favorable reservoirs gradually thicken from north to south, exhibiting a banded distribution with thickness primarily between 15 and 30 m. This study provides a reliable geological basis for subsequent well location deployment in the target area and serves as a technical reference for reservoir prediction in similar blocks.