The process optimization of p-xylene separation from p-/o-xylene binary mixture through silicalite-1 membrane using statistical design of experiments (DoE) is reported in the present study. The silicalite-1 membrane was synthesized and characterized using different analytical techniques. The effect of three important process variables, temperature (150-250 degrees C), p-xylene feed partial pressure (0.04-0.50 kPa) and p-xylene feed composition (0.20-0.80) on the separation performance of the membrane was studied. The response surface methodology (RSM) coupled with central composite design (CCD) was used to develop three models to correlate the effect of process variables to three responses: (i) p-xylene flux, (ii) o-xylene flux and (iii) p-/o-xylene separation factor. The most influential factor on each of the response was identified using the analysis of variance (ANOVA). The interaction between the three variables was systematically investigated based on three-dimensional response surface plots. The optimum operating condition for the process was determined by setting the optimization criteria to maximize the p-xylene flux and p-/o-xylene separation factor, and to minimize the o-xylene flux. The optimum p-xylene flux of 3.83 x 10(-6) mol/m(2) s and p-/o-xylene separation factor of 46 were obtained at a temperature of 198 degrees C, p-xylene feed partial pressure of 0.15 kPa and p-xylene feed composition of 0.80. The simulated values obtained from the statistical model were in agreement with the experimental results within an average error of +/-2.70%. The mass transport of xylene isomers and its separation in the silicalite-1 membrane was related with the characteristics of the membrane. (C) 2009 Elsevier B.V. All rights reserved.