Contamination of rice bran with the husk adversely affects the efficiency of an oil extraction process. The efficient separation of husk from the rice bran prior to an oil extraction process is of great benefit. In the present work, a three-factor-three-level Box-Behnken design of experiments combining Response Surface Methodology (RSM) is being employed to optimize the air classifier parameters. Three independent variables (i.e., wheel speed ranging from 800 to 1200 RPM, guide vane angle ranging from 13 to 65 degrees, and the feed rate ranging from 12 to 24 kg/h) are consecutively coded as x(1), x(2), and x(3) at three levels (i.e., -1, 0, and 1). Models were developed to demonstrate the effect of each parameter and their interaction effects on the responses. The second-order polynomial regression equations were derived from the data to predict the classifier cut size and the percentage of oil content. The significance of independent variables and their interactions are being tested by the analysis of variance at 95% confidence level ( = 0.05). The predicted values of responses obtained from the models were in good agreement with the experimental values. For the improvement of the oil extraction process, a pre-concentration methodology by a circulating air classifier is demonstrated and recommended to the industry for optimal separation of the rice bran/husk.