Hybrid distillation/melt crystallization processes are widely used for the separation of isomer mixtures. The design of a hybrid separation processes is not a trivial task because of several structural and operational degrees of freedom. In this article, a new three-step design approach is proposed to address the above mentioned challenges. In the first step, process alternatives are generated by heuristic rules. In the second step, the generated process alternatives are optimized with respect to their energy consumption by using shortcut methods. In the third step, the most promising alternatives are rigorously optimized with respect to their total annualized costs, whereas the resulting mixed-integer nonlinear programming (MINLP) problem is solved using a modified generalized benders decomposition (GBD) algorithm to take the nonconvexities into account. The feasibility of this design approach is demonstrated by design of a hybrid distillation/melt crystallization process for separation of a ternary isomer mixture. (C) 2008 American Institute of Chemical Engineers AIChE J, 54: 2925-2942, 2008