The synthesis of complex processes, those involving reaction and separation sections with recycle streams and with heat integration, is a computationally difficult task. It is also a task which is difficult to formulate. This paper describes a discrete programming approach, using dynamic programming techniques, for the automated synthesis of such processes. The problem formulation is easy and results are generated quickly, encouraging experimentation by the user, a desirable objective in the early stages of design. The use of dynamic programming enables the generation of multiple solutions, each of which is near optimal, providing the user with a small set of good alternatives. This small set of alternatives can then be analysed using more rigorous continuous optimization techniques, fine tuning some of the continuous parameters in the solutions.