Abstract: The vacuum preloading method is widely used in the improvement of large area with deep soft cohesive soils in the coastal plains, which can accelerate the drainage consolidation of the soft clay and improve the shear strength. Based on the in-situ and laboratory tests before and after the vacuum preloading improvement, the physical and mechanical indices (moisture content and void ratio) as well as strength indies (shear strength and penetration resistance) of the soft clay before and after the reinforcement were compared and analyzed by mathematical statistics approach. Based on the difficulties that frequently encountered in the engineering practice of using the conventional analytical approach, such as, small reliability indices, and large deviation from the measure conditions, this paper presents the use of random field approach to obtain the correlation distance and variance reduction function values of soil indices after soft clay improvement in this region. The results are applied to the foundation soil reliability evaluation. The results show that the soil indies of the soft clay after the vacuum preloading improvement are consistent with the values in the random field theory model, and the reliability indices of the soil bearing capacity considering the correlation of soil indies is significantly improved, which is more in line with the engineering practice.