This paper shows how the large-scale discrete version of the stochastic linear-programming problem with simple recourse can be transformed into a sequence of problems with greatly reduced dimensions for ready solution by the primal-dual approach of the dualplex method. Algorithmic details and initial computational results are discussed.