This paper presents the investment planning model, which has been developed to rigorously valorize the large hydroelectric potential resources of Cameroon. Based on Benders technique, the decomposition approach proposed allows each component of the problem (investment and operation) to be conveniently modelled and solved by adapted methods. Furthermore, the relative small size of the actual system, the long experience of the manual generation management in use over a long time by the national electricity company (Sonel), and the fundamental data of the problem are exploited as the basis of the approach, to define a more detailed and adapted model. Formulated as a multistage mixed linear programming problem, the decomposition procedure of the Benders technique is applied to isolate from the problem so modelled its two main subproblems (investment and operation). The investment subproblem is modelled as a multistage mixed linear programming problem and solved by a branch and bound technique. The operation subproblem is represented as a multistage stochastic linear problem and solved by a revised simplex-based method. Subsequently, a solution algorithm for the whole problem, defined by a finite iterative process where the subproblems are alternatively solved, is derived from the co-ordination procedure of Benders technique. Case studies with the southern interconnected network of the Sonel system are presented and analysed. Copyright (C) 2001 John Wiley & Sons, Ltd.