This paper deals with the problem of evaluating and optimizing the expected stochastic flexibility in multiproduct batch plants. This measure, which integrates flexibility and reliability considerations, reflects the probability of meeting demands. It is assumed that uncertainties in the product demands are given in terms of normal distribution functions and that expected failure rates of the equipment are given in the form of discrete probabilities. Based on recent work by the authors (Straub and Grossmann, Computers chem. Engng 14, 967, 1990), efficient computational methods are proposed to evaluate the expected stochastic flexibility which exploit the structure of multiproduct batch plants. It is also shown that these methods can be incorporated in the optimization for determining the sizes and parallel equipment that maximize the expected stochastic flexibility under a capital investment constraint. By varying the specification of the latter, trade-offs can be established between investment cost and expected stochastic flexibility. Based on these optimization models, an approximate procedure is also presented for determining designs with maximum expected profit.