In many rural communities, groundwater is used to meet the water demand of the community and for the irrigation of cultivating areas. The quality of groundwater can be adversely affected by agricultural activities and finally groundwater quality may become unsuitable for human consumption and irrigation, as in the Harran Plain. Hence, monitoring groundwater quality by cost-effective techniques is necessary, as especially unconfined aquifers are vulnerable to contamination. This study presents an artificial neural network model predicting sodium adsorption ratio (SAR) and sulfate concentration in the unconfined aquifer of the Harran Plain. Samples from 24 observation wells were analyzed monthly for 1 year. Electrical conductivity, pH, groundwater level, temperature, total hardness and chloride were used as input parameters in the predictions. The best back-propagation (BP) algorithm and neuron numbers were determined for the optimization of the model architecture. The Levenberg-Marquardt algorithm was selected as the best of 12 BP algorithms and optimal neuron number was determined as 20 for both parameters. The model tracked the experimental data very closely both for SAR (R = 0.96) and sulfate (R = 0.98). Hence, it is possible to manage groundwater resources in a more cost-effective and easier way with the proposed model application.