A multi-cell model was developed to analyze the behavior of a simulated moving bed process for adsorptive para-xylene separation from other xylene isomers. A novel technology for a semi-batch mode adsorption experiment was developed and used for fast and accurate data collection. Interaction parameters between different species for a multi-component extended Langmuir isotherm were estimated from single and multi-component adsorption experiments and implemented into the model. The parameters such as porosities, particle density and mass transfer coefficients were obtained from adsorbent analysis and commercial plant operation. To resolve the problem of high dimensionality, a cell-by-cell approach was proposed to solve the model. The recovery and purity of para-xylene as well as the concentration profile calculated from the model were in good agreement with the actual data. The effects of channeling and feed composition change were simulated, and they turned out to be physically meaningful. The simulation model will be used for operation condition optimization, trouble shooting, and productivity enhancement including a configuration change.