Electrical demand prediction is an important issue in power system planning. An improved electricity prediction method is put forward, which looks out the electricity consumption level in long term future first, then utilizes Logistic curve to fit the electricity consumption growth with saturation process. The future developing scenario constrains are taken into account in the parameter identification step of Logistic curve, which makes the curve beable to not only reflect the historical changing trend but also be in accordance with the future scenario. This method is applied to predict the electricity consumption of China and Guangxi Province in case studies respectively, and shows feasible and significant to other power demand prediction tasks.