ANNs have been used as simplified black box models in several water resources problems. Black-box flood prediction models can be thought of as model mappings to input-output datasets. The multilayer perceptron (MLP) network (Rumelhart et al., 1986) is usually used for model mapping problems and consists of the input layer, one or more hidden layers of computational nodes (neurons), and an output layer of computational nodes . In a feedforward structure, data flows only in one direction. A neuron in a particular layer receives all the inputs from the previous layer and transmits the values ​​to subsequent layers of the processing element. The schematic representation of the feedforward MLP network designed according to the Muskingum equation (Eq. 4) is shown in Fig. 1. It has an input layer, a hidden layer and an output layer, respectively. Each node of a layer is connected to all the nodes of the next layer and the neurons of the same layer are not connected to each other. The data transferred from one neuron to another across connections is controlled by weights. The sum of the product of the inputs (x) and their weights (w) minus the bias (b) leads to a network as follows: net = ∑▒〖x_i .w_i 〗 - b (12)Thus, the output of a neuron, f (net) is decided by an activation function that determines a response of the node to the input signal it receives. To generate an output vector, a training (learning) process is used to find optimal weight matrices and bias vectors that minimize a predetermined error function (E) which is the sum of the squares of the error given as follows.E =∑_(p =1 )^p▒∑_(i=1)^N▒〖 (y_i-t_i)〗^2 (13)Here, ti and yi represent the objective (observed outflow, O_(t+∆ t)) and output ( expected runoff... halfway through the document... to its important role in the country's social economy. Every year, in one place or another, severe floods hit the Chindwin Basin at due to the high intensity of rainfall during the southwest monsoon. Since 1965, flood events in the Chindwin basin have been the highest in the months of July and August, contributing to 72% of the total number of floods in the basin This study reported floods of two hydrometric stations namely Homalin and Mawlaik were taken into consideration the length of the stretch from upstream (Homalin) to downstream (Mawlaik) is 114 km. According to the topography of the basin, the slope of the section is about 0.0003, which is relatively mild. A flood wave is subject to translational and reservoir actions during its passage through natural rivers and hydrological methods are appropriately used in case the reservoir action dominates the translational action for rivers with a gentle slope (Das, 2009).