A general procedure for robust design of simulated moving-bed (SMB) processes under flow rate uncertainty is presented. The best solution is chosen only among candidate solutions that are robust feasible, that is, remain feasible for all flow rate perturbations from the uncertainty set. This gives rise to a robust approach to optimal SMB design in which the nominal problem is replaced by a worst case problem. Computational tractability is ensured by formulating the robust problem with only the vertices of the uncertainty region that most adversely affect the raffinate and extract purities. The nominal optimization problem and its robust counterpart are formulated with a single-column analog model, and a full-discretization approach for steady periodic dynamics. The resulting nonlinear programming problems are solved by an efficient interior-point solver. The procedure is successfully employed to find robust operating conditions for the linear separation of two nucleosides by standard SMB, asynchronous port switching (Varicol), and cyclic flow rate modulation (PowerFeed). Our results show that PowerFeed is the most efficient operating scheme for the separation under study, although Varicol also clearly outperforms the standard SMB process. However, it is also shown that Varicol productivity is the least affected when the process is rendered robust, whereas that of PowerFeed is the most penalized. (c) 2007 American Institute of Chemical Engineers.