This paper aims to forecast the space distribution of International Tourist Hotel. The Neural Network Model is applied to find out the changes of the number of rooms in ITH from 1980 to 1998, and the eleven related phase items is applied to proceed such forecast. First, it is forecasted in zero-hidden layer one-hidden layer and two-hidden layer respectively, then selects a model with both better observation and forecast values. In addition, the forecasts of the changes between Community Tourist Hotel and International Tourist Hotel in the sections for north, central, south, and east is conducted to evaluate the future development.