This article addresses the design of chemical supply chains (SCs) under demand uncertainty. Given a design horizon consisting of several time periods in which expected but uncertain product demands materialize, the objective is to select the design that maximizes the expected profit. To tackle this problem, a multistage stochastic formulation is derived wherein the design decisions are made irrespective of the specific realization of the uncertain parameters. The resulting model is, in general, computationally intensive or even intractable, as the number of variables and equations is proportional to the number of possible nodes on the scenario tree, which explodes exponentially with the number of stages ( periods). In view of this, a decomposition technique is introduced aiming at the objective of overcoming the numerical difficulties associated with the underlying large-scale stochastic mixed- integer linear problem (MILP). The resulting strategy combines genetic algorithms (GAs) and mathematical programming tools. Computational results indicate that the proposed approximation method provides solutions which are within a few percent of the optimal ones and also reduces the computation time required by the rigorous mathematical model.