The treasury yield curve is an important reference in asset pricing,macro- control and monetary- policy formulation. In this paper,we extract three principal components that describe the large proportion of the movement of the treasury yield curve,combining various macro- economic variables,to make regression analysis of the treasury yield curve. Historical data analysis indicates that the treasury yield can be well fitted,and that the movement of the treasury yield can be pre- determined to some extent. We also adopt the stepwise regression method,searching for the significant variables with co- integration relation,and find that the significant variables can be useful in the prediction of the changes of the treasury yield curve.