The water film thickness of cement ground limestone paste was calculated by the total specific surface area and packing density of the particles. The relationship between the water film thickness and the total specific surface area and packing density of the particles was investigated. The yield stress of the paste was measured and the model of relationship between water film thickness and yield stress was established based on BP neural network. The results show that the water film thickness of cement ground limestone paste depends on the ratio of packing density to total specific surface area and increases with the increase of the ratio. The change of water film thickness is significantly correlated to the relative increase rate of packing density and total specific surface area. When the increase rate of packing density is greater than the increase rate of total specific surface area of particles (Δ>0), the water film thickness increases and vice versa. The Peasron correlation coefficient between calculated values and measured values of the model of yield stress is 0955, which means the model has high accuracy.