The amount of sediments transported by headwater rivers plays a crucial role in planning of water resources. Most widely used methods of estimating the sediment in rivers are the empirical methods, and in India, the rating curve technique is most popular. The present study is focused on the application of artificial neural network (ANN) technique for sediment-discharge modelling of a headwater river. For ANN development, daily discharge and suspended sediment concentration data of Subansiri River (an eastern Himalayan river) in India have been used. Rating curves have also been developed with similar data, and comparison of the two techniques has been carried out. It has been observed that the estimates of suspended sediment concentration obtained by ANNs compared to the rating curve technique were much closer to the observed values. © 2017 Springer Science+Business Media Singapore.