Abstract: The key to improve the accuracy of height anomaly fitting is to select the appropriate fitting model. Based on the GNSS and leveling data in Xinhua County, Hunan Province,this paper studies and evaluates the fitting accuracy of height anomaly by using inverse distance weighting method,polynomial interpolation method, radial basis function method(RBF),Kriging interpolation method and other surface fitting methods.The results show that the radial basis function interpolation has the highest interpolation accuracy,and the error is ±0.0158 m.The local polynomial interpolation method owns the highest extrapolation accuracy,in which the error is ±0.0104 m.That means the local polynomial interpolation method has the highest accuracy in county scale elevation fitting.The research conclusion of this paper can be important reference significance for the selection of GNSS Height Fitting Method in small areas.