The invention puts forward an urban tidal river reach ecological water demand prediction method based on a neural network, and belongs to the field of ecological hydrology. The method comprises the following steps that: firstly, carrying out statistics on the monthly seawater salinity and sea-entering water quantity of the estuary of the urban tidal river reach in multiple years to obtain the monthly salt balance water demand of the river reach and establish the monthly salt balance water demand prediction network of the urban tidal river reach; then, independently obtaining the monthly average rain capacity, average sea surface temperature, total sewage discharge amount and ecological water demand of the river reach, and establishing the monthly ecological water demand prediction network of the urban tidal river reach; and independently obtaining the prediction value of the monthly total sewage discharge amount, the monthly sea-entering water quality prediction value and the monthly average rain capacity prediction value of the river reach in certain year in the future, inputting into the ecological water demand prediction network, and outputting the ecological water demand prediction result of the urban tidal river reach under different conditions. By use of the method, the special properties of the urban tidal river reach are comprehensively considered, a practical situation is conformed, the change situation of the river reach ecological water demand in the future is predicted, and a reference is provided for a decision department.