We consider two-stage stochastic programming problems with integer rec
ourse. The L-shaped method of stochastic linear programming is general
ized to these problems by using generalized Benders decomposition. Non
linear feasibility and optimality cuts are determined via general dual
ity theory and can be generated when the second stage problem is solve
d by standard techniques. Finite convergence of the method is establis
hed when Gomory's fractional cutting plane algorithm or a branch-and-b
ound algorithm is applied. (C) 1998 The Mathematical Programming Socie
ty, Inc. Published by Elsevier Science B.V.