Water is one of the most important natural resources for living life, and its rapid population growth, unconscious use of water in agriculture and industry and its pollution may lead to difficult situations. Water quality is important not only for human life but also for all other living things. One of the most important parameters regarding water pollution is dissolved oxygen. Dissolved oxygen is vital to aquatic life. Artificial neural networks have uses for water quality parameters along with other applications in water management. Significant gains can be achieved by modeling water quality parameters with artificial neural networks. In this study, the modeling of the dissolved oxygen value with the water quality data measured in the months between 1995−2014 belonging to the Sakarya Basin monitoring stations was made. Temperature, electrical conductivity, pH, biological oxygen demand, and months are used as input data in model. Also, dissolved oxygen ise used output data. There are 1388 data in total. 80% of the total data (1107 data) was used for training, 20% (281 data) was used as test data. In this study, Root Mean Square Error, Mean Absolute Error, Mean Absolute Relative Error, and Nash−Sutcliffe Model Efficiency Coefficient criteria were used to measure the prediction success of the models. As a result of the study, the error criteria are reasonable and its predictive power is high