The accurate prediction of water demand has great significance for the development of city. The water demand is affected by multiple factors and most of these factors have strong relevance. Based on the calculation steps of the principal component analysis, and introduced an instance (water consumption data of Jinhua city from 2000to 2010), the regression model of predicting water demand was established. The results showed that the model was applied to forecast water demand, and the conclusion agreed quite well with the local actual conditions, the model fitting degree and prediction accuracy were ideal.