Rivers are ecosystems that are significantly affected by environmental pollution. For this reason, the management of rivers for sustainable water management needs to be well managed and its pollution must be well identified and monitored. In this study, biological oxygen demand (BOD), chemical oxygen demand (COD), suspended solids (SS), pH, conductivity (CE) and temperature (T) values were examined in five locations between December 2012 and December 2013 in Bartin River. Then multiple linear regression (MLR), Radial Basis Neural Network (RBANN), Multilayer Perceptron Neural Networks (MLP) models were applied for water quality forecasting. In these models, BOD value was estimated by using T, pH, COD, SS, CE parameters as input data. Forty-one measurement data belonging to the locations were used in the training and the other 18 measurement data were used in the test process. According to the obtained results, Artificial Neural Network (ANN) models have shown better results than multiple linear regression model. Compared to the established models, the best performance values were achieved with a radial based artificial neural network model. In this model MAE, RMSE and R2 values obtained 0.998, 1.230 and 0.890 respectively. According to the results of the present research the most successful estimation by ANN models was achieved for the monthly BOD values in Bartin River. © 2018 Elsevier B.V., All rights reserved.