This paper describes the application of an industrial real-time optimization package (ROMeo(1)) to the nonlinear model predictive control (NLMPC) of a simulated polymer grade transition. The NLMPC algorithm is formulated using orthogonal collocation to integrate the model equations, and sequential quadratic programming to solve the resulting nonlinear programming problem. A receding horizon estimation scheme and, separately, a Luenberger observer are designed to reconstruct unmeasured states. The resulting algorithm is demonstrated on two simulated polymerization case studies: (i) continuous methyl methacrylate and (ii) gas-phase polyethylene. (C) 2003 Elsevier Ltd. All rights reserved.