Abstract: Based on the neural network model of time series and the periodical changing characteristic of the groundwater level in dry region,a seasonal artificial neural network forecast model is set up.At the same time the data smoothing process is carried out aimed at the different fluctuation during different period before these data are used for training.The prediction results indicate that the model is reasonable,its accu racy is better,and has some practical value when being used in the dynamic groundwater level analysis.