Abstract: Three functional models,polynomial,spectral analysis,and modified AR model,are fitted and compared in fitting and predicting the clock differences based on the data series derived from two-way satellite time and frequency transfer.A robust equivalent weight is applied,which controls the significant of outlying observations.Some conclusions show that the prediction precision of robust estimation is better than that of LS.The prediction precision calculated from smoothed observations is higher than that calculated from sampling observations.As a count of the obvious period variations in the time series,the prediction result of polynomial model is uncertainty.The prediction precision of spectral analysis is very low,but the main period variations can be determined.The 6-hour prediction RMS is 1 ns or so,when modified AR model is used.