In this study, Taguchi-neural approach was used to find the real optimum condition for the forward osmosis (FO) groundwater desalination. The feed solution velocity, draw solution velocity, feed solution temperature, and draw solution temperature (denoted as A, B, C, and D, respectively) were chosen as operating parameters, whereas maximum reverse solute flux selectivity (RSFS), J(v)/J(s), was chosen as the response. In the first step, the Taguchi method was applied to evaluate the primary optimal condition. The effect of parameters on both an active layer facing feed solution (AL-FS) and an active layer facing draw solution (AL-DS) orientations were evolved according to L16 (4(4)) orthogonal array. The primary optimal condition for AL-FS, i.e., A = 40 cm/min, B = 30 cm/min, C = 25 degrees C, D = 55 degrees C, and for AL-DS, A = 20 cm/min, B =40 cm/min, degrees C = 25 degrees C, D =55 degrees C were obtained. Analysis of variance (ANOVA) was used to recognize the important parameters that could affect FO quality characteristics (J(v)/J(s)). Statistically, draw solution temperature (55 degrees C) was found to be the most important parameters for quality characteristics (maximum J(v)/J(s)). In the second step, all experimental results of AL-FS and AL-DS were used to train the neural network. The trained neural networks were used to find real optimum parameter levels. For AL-FS, A = 34 cm/min, B = 10 cm/min, C = 35 degrees C, D =50 degrees C, and for AL-DS, A= 18 cm/min, B = 33 cm/min, C= 35 degrees C, D =48 degrees C were found to be the real optimum parameters. This study shows the possibility of optimizing the FO process proficiently using the Taguchi-neural network approach. (c) 2016 The Institution of Chemical Engineers. Published by Elsevier B.V. All rights reserved.