To study the relationship between matrix suction and conductivity in unsaturated granite residual soil and realize the matrix suction prediction of soil slope based on conductivity, laboratory and field tests are carried out on undisturbed soil at different depths of the Yandou village landslide in Sanming City, Fujian Province, China. Through physical and chemical property analysis, soil-water characteristic curves and electric parameter matrix suction prediction models for unsaturated granite residual soil at different depths of the target area are obtained. Based on the proposed model, the matrix suction distribution of on-site soil slope is predicted and the dynamic response law under the influence of artificial rainfall is studied. The results show that: (1) The transverse conductivity, average structure factor, average shape factor, and anisotropy coefficient of unsaturated soil are related to the soil saturation degree. By considering the above parameters, the comprehensive structure parameter Re is introduced and its functional relationship with matrix suction is established. (2) Under artificial simulated rainfall, the saturation, hysteresis of the conductivity parameters, and matrix suction response of the slope occurs, which is controlled by soil depth, permeability and rainfall intensity. The matrix suction is distributed in layers on the profile and its recovery rate is slower than saturation. The suction contour map shows a parabola shape with the opening downward. (3) The relationship between the conductivity parameters of the residual soil slope and matrix suction is further revealed and a new method to indirectly measure matrix suction is proposed. Its feasibility is verified based on field tests, which is of great significance to landslide monitoring and early warning.