This study describes an application of model-based control techniques to a liquid-liquid extraction process. The control design is based on a reduced order model of the process obtained by system identification. A multi-loop decentralized control system, in the framework of internal model control (IMC), is initially selected to control the 2 x 2 control loops of the extraction column. The interaction of loops is then investigated to explore the feasibility of such multi-loop control structure. A centralized multivariable control system is synthesized with a model predictive control (MPC) technique. The simulation results demonstrate good servo and regulatory characteristics for both control system designs. However, the MPC control of extraction proved to be superior to the IMC one in terms of speed of response, stability, robustness, and loops interaction. MPC has also shown ability to handle control operation under input variables constraints, which has significant practical importance. (C) 2004 Published by Elsevier B.V.