Abstract: To raise the forecast precision of railway freight ton-kilometers(RFTK),seven factors reflecting the supply capacity of railway transportation were selected out based on a qualitative analysis and the grey relevancy degree theory,and a semi-parametric regression model was established.The partial least-squares regression method was used to process the multicollinearity of variables,i.e.,the seven factors,and the non-parameter method was applied to express qualitative factors.The research result shows that with the established semi-parametric regression model,the least relative error of the forecast result for RFTK is only 1.7%,and compared with the linear regression model and the grey forecast model,the semi-parametric regression model has a better effect and higher precision for the forecast of railway freight ton-kilometers.