The purpose of this paper is to evaluate a heuristic procedure for resource-constrained project scheduling. The procedure is a representative of a class of multi-pass procedures based on problem decomposition and is applicable to single or multiple project networks. Solutions found were comparable to those obtained with a branch-and-bound optimization algorithm while computing times were significantly shorter. Analysis based on the sample indicates the procedure is much less sensitive to problem size than the branch and bound algorithm. The method of decomposition and the evaluation procedures are described in detail.