This paper presents the use of artificial neural network (ANN) to develop a model for predicting rejection rate (R-o) of single salt (NaCl) by nanofiltration based on experimental datasets. The rejection rates of NaCl were obtained when operating conditions, such as feed pressure (Delta P) and cross flow velocity (V), varied along with different physicochemical properties of feed water like salt and dye concentrations, and pH. In the modeling work, sensitivity analyses were performed to identify relative impact of each parameter and to find the best combination of input parameters in the ANN model. The optimal network architecture was developed through trial and error approach. Model predictions in each trial were compared with experimental results based on statistical evaluation such as root mean square error, mean absolute error, and coefficient of determination (R-2). Optimal network architecture was determined as one hidden layer with 25 neurons using Levenberg-Marquardt (trainlm) back-propagation algorithm. In this architecture, tangent sigmoid (tansig) in hidden layer and linear (purelin) in output layer was also used as transfer functions. The results showed that the developed ANN model predictions and experimental data matched well and the model can be employed successfully for the prediction of the R-o.