On the basis of the gray system theory, we weaken the original data of water demand by using several methods, then conduct comparative analysis on forecasting results. Analysis shows that the result obtained from the combination of second-order weakening operator with the AGO can get the highest accuracy. Finally, we validate the feasibility and effectiveness of the model which is applied to forecasting water demand in Beibei, Chongqing. The forecasted result is that: water demand of Beibei District is 245 million m_³ in 2015, and 269 million m_³ for the year of 2020.