This study took the Changzhou site as an example, and collected the historical data of meteorological observation during 2000—2017 and the historical data of weather forecast during 2011—2015. We calibrated the Hargreaves-Samani model, the Multiple Regression model and the Fourier analysis model by using the ET₀ calculated by historical meteorological data based on the FAO56-PM formula, and applied these three methods to the daily ET₀ prediction based on the weather forecast for the next one to seven days during the period form April 21, 2016 to October 24, 2017. The results showed that the MR models had the highest prediction accuracy overall with the lowest δ_MAE(0. 751 mm/d)and the highest δ_AAC(87%)during the forecast periods of 1 d, 4 d, and 7 d, except that δ_RMSE was slightly higher than the HAR model during the forecast period of 3 d. In view of the fact that the accuracy of the weather forecast decreases with the increase of forecast period, it is suggested that the forecast period of Changzhou station should not be longer than three days. Further comparison on the accuracy of seasonal scale showed that the prediction accuracy of the MR model during the forecast periods from 1 d to 3 d in each season is higher than those of HAR model and FA model. Overall, the MR model has the highest prediction accuracy and should be used for ET₀ prediction at the Changzhou site.