The operation of chemical processes is inherently subject to uncertainty. Traditionally, uncertainties have been accounted for in system design by discretizing the uncertainty space and considering the resulting ensemble of scenarios in solving the design optimization problem. Scenario-based approaches are computationally demanding and can rapidly become intractable. We propose identification-based optimization (IBO) as a novel framework for the optimal design of dynamical systems under uncertainty. Our method originates in nonlinear system identification theory, and is predicated on representing uncertain variables as pseudo-random multi-level signals (PRMSs), which are imposed on the system model during each time integration step of a dynamic optimization. The uncertainty space is thus efficiently sampled without using computationally expensive scenario sets. We establish a procedure for generating PRMSs for uncertain variables based on their probability density functions. The computational benefits of IBO are illustrated through comparative case studies. (C) 2014 Elsevier Ltd. All rights reserved.