In the context of the recently developed equation-free modelling framework, we present a computer assisted approach to locating approximate coarse optimal switching policies between stationary states of chemically reacting systems described by microscopic/stochastic evolution rules. The coarse lime-stepper constitutes a bridge between the underlying stochastic simulation and traditional, continuum numerical optimization techniques formulated in discrete time. The approach is illustrated on a simple CO oxidation on Pt catalytic surface reaction model, implemented through the Gillespie stochastic simulation algorithm. The objective sought is to switch between two coexisting stable stationary states by minimal manipulation of a macroscopic system parameter. Copyright (c) 2005 John Wiley & Sons, Ltd.