Abstract: Gas-liquid spontaneous imbibition in microchannels is a widely occurring physical phenomenon in nature and many industrial fields. The dynamic contact angle is the key factor affecting the whole gas-liquid imbibition process. In this work, we use a modified pseudopotential multiphase flow lattice Boltzmann method (LBM) to capture the real-time contact angle during gas-liquid spontaneous imbibition in microchannels and analyze the dynamic characteristics of the contact angle and its effects on the imbibition length. Firstly, we coupled the Peng-Robinson (PR) equation of state to the original pseudopotential multiphase flow LBM, improved the fluid-fluid interaction force and fluid-solid interaction force formats, and added the external forces to the LBM framework by using the exact difference method. Then, the accuracy of the model was verified by calibrating the thermodynamic consistency of the model and simulating interfacial phenomena such as interfacial tension and static equilibrium contact angles. Finally, based on the established simulation method, the spontaneous gas-liquid percolation process in the microchannel is simulated in the horizontal direction. The results show that the contact angle in the imbibition process is dynamic and varies greatly in the early stage of imbibition due to the inertia force. With the further increase of the imbibition distance, it gradually decreases and tends to the static equilibrium contact angle. The contact angle in the imbibition process is related to the microchannel size and the static contact angle. As the width of the microchannel increases, the difference between the dynamic contact angle and the static contact angle in real-time increases; as the static contact angle increases, the difference between the dynamic contact angle and the static contact angle in real-time increases. In addition, the Lucas-Washburn (LW) equation, which ignores the dynamic contact angle, predicts the position of the meniscus is different from the simulated results. The real-time dynamic contact angle data obtained from the simulations can be directly applied to correct the LW equation, and the corrected LW equation predicts the position of the meniscus in general agreement with the simulated results.