The performance of a cyclone is generally assessed using the pressure drop and the collection efficiency of the cyclone. In the present paper, a multi-objective optimization of a classic Stairmand cyclone separator is executed using the response surface methodology (RSM), combined with computational fluid dynamics (CFD) techniques for minimizing the pressure drop and maximizing the collection efficiency. Ten cyclone geometrical factors are considered in this work. Three of them are studied using the RSM according to the results of the preceding screening experiments. The second-order response surface models for each response are successfully carried out using the central composite design (CCD) in the RSM. The desirability function approach is used to optimize the geometrical factors of the cyclone. In comparison to the reference model, the optimal cyclone model decreases the pressure drop by 20.70% and the cut-off size by 75.38%. The accuracies of the response surface models are confirmed and the correctness of the optimal cyclone model is also validated using CFD; the results indicate excellent performance and reliability of the response surface model and the RSM optimization result. According to the analyses of the obtained flow fields, there are several reasons why decreases in the cut-off size with a lower pressure drop are found in the optimal model. The primary reasons for improvements are optimized values of relevant geometrical factors, the production of a distinct, undisturbed downward flow, an increase in the tangential velocity at the near wall region, and a decrease in the peak tangential velocity. (C)2017 Elsevier B.V. All rights reserved.