Abstract: To address the uncertainty in evaluating the lateral bearing performance of offshore wind turbine monopile foundations, a two-dimensional random finite element model considering the spatial variability of soil undrained shear strength was developed in this study. The random field was generated using the Karhunen-Loève expansion combined with a Gaussian autocorrelation function, and the probabilistic responses of the monopile foundation were obtained through Monte Carlo simulation. Based on a series of representative loading cases, the effects of the coefficient of variation (COV) and correlation length on the ultimate lateral capacity, pile bending moment, and shear force were systematically investigated. The results show that COV is the dominant factor governing the dispersion of pile responses. As COV increases, the uncertainty of the ultimate lateral capacity becomes significantly more pronounced; under high-COV conditions, its mean value tends to decrease, while the range of extreme responses expands markedly. In contrast, the correlation length has little influence on the mean response, but significantly amplifies the fluctuations and standard deviation of structural responses. Moreover, pile shear force is found to be more sensitive to soil spatial variability than pile bending moment. Under highly variable conditions, the mean shear force is slightly higher than that predicted by deterministic analysis, accompanied by a substantial increase in dispersion. These findings indicate that conventional deterministic analysis may underestimate the risk associated with shear response under complex geological conditions. The proposed approach provides a useful theoretical basis for the reliability-based design of offshore wind turbine monopile foundations.