The objective of the investigation is to carry out the flow routing in the natural channel of the experimental basin of the Ichu river, by means of the Artificial Intelligence technique of ANNs (Artificial Neural Networks). Generally, hydrological and hydraulic methods require different parameters of the river channel, while the ANNs method simplifies the amount of data. The study area is located in the experimental basin of the Ichu river, upstream of the city of Huancavelica in an area of 607 km 2 . A calibrated and validated model of the rain-runoff process was developed, with data recorded in 6 automatic meteorological stations (rainfall) and one hydrological station (runoff). The model HEC-1 was used to model the rain-runoff process and the Muskingun-Cunge method for the flood rounting, generating historical records for 5 stretches of the Ichu riverbed and obtaining 39 maximum historical records in the 2016 periods and 2017. The model obtained values of Nash-Sutcliffe efficiency coefficients (E) equal to 0.851 and 0.828 for the calibration and validation stage, respectively. The ANNs were built with different architectures to train and obtain the architecture that best fits the historical phenomena. Finally, the architecture 1-5-1 presented a better fit, whose statistical E was values of 0.881 and 0.859 in the training and validation stage respectively.