The cotton/insect-pest agroecosystem is extremely complex temporally and spatially. To adequately manage such a system, aspects of this complexity must be considered. New techniques in computer science, derived from research in artificial intelligence, make it possible to represent such system detail. WHIMS, a knowledge-based (expert) system for cotton pest management, is being developed using such techniques. WHIMS will consist of a model of the cotton/pest complex coupled with a model of the management decision domain. The cotton/pest complex model will have an object-oriented framework and, through the incorporation of qualitative modeling techniques from artificial intelligence, be capable of representing physical and biotic processes qualitatively as well as quantitatively. The decision domain model is being developed using Bayesian Belief Network technology. This will incorporate the uncertainty inherent in our knowledge of the pest-management domain.