This document summarizes a study that uses a wavelet-neural network (WLNN) conjunction model for river flow forecasting of the Brahmaputra River in India. The model decomposes river discharge time series data into multiresolution time series using discrete wavelet transforms. These decomposed time series are then used as inputs to an artificial neural network (ANN) to forecast river flows at different lead times. The results of the WLNN model are compared to those of a single ANN model. The WLNN model is found to provide more accurate and consistent predictions than the ANN model alone due to its use of multiresolution time series data as inputs.