A two-layer mathematical model proposed by Tong et al. (2010) was used to predict soluble chemical transfer from soil into surface runoff with ponded water on the soil surface. Infiltration-related incomplete mixing parameter γ and runoff-related incomplete mixing parameter α in the analytical solution of the Tong et al. (2010) model were assumed to be constant. In this study, different laboratory experimental data of soluble chemical concentration in surface runoff from initially unsaturated and saturated soils were used to identify the variables γ and α based on the analytical solution of the model. The values of γ and α without occurrence of surface runoff were constant and equal to their values at the moment when the surface runoff started. It was determined from the results that γ decreases with the increase of the ponded water depth, and when the initial volumetric water content is closer to the saturated water content, there is less variation of parameter γ after the occurrence of surface runoff. As infiltration increases, the soluble chemical concentration in surface runoff decreases. The values of parameter α range from 0 to 1 for the fine loam and sand under the controlled infiltration conditions, while it can increase to a very large value, greater than 1, for the sand under the restrained infiltration conditions, and the analytical solution of the model is not valid for experimental soil without any infiltration if α is expected to be less than or equal to 1. The soluble chemical concentrations predicted from the model with variable incomplete mixing parameters γ and α are more accurate than those from the model with constant γ and α values.