The ability to predict larval phenology for the gypsy moth,Lymantria dispar(L), is of interest for a variety of reasons, ranging from basic ecological research applications to the applied objective of improving timing of spray applications. Problems with the development of an appropriate model to predict larval phenology for this pest are exacerbated by the large, and often complex, landscapes that must be considered in control programs, and by the gypsy moth's extensive geographic distribution. The goal of research discussed in this paper is to provide a flexible modeling system that can be used to construct a phenology model that is well suited for a particular application. To accomplish this goal, parameters for a variety of developmental models from an existing data base were estimated. Basic modeling assumptions, such as the instar independence of developmental rates, were tested. Parameter estimation for the set of reasonable models resulted in a large number of potential phenology models. To assist in the development of an effective predictive tool, a modeling system was designed for assembling and implementing selected models. Finally, several applications were discussed that illustrate the rationale for choosing the most appropriate model for a particular application. Our primary objective is to provide a convenient modeling framework for field testing and applying gypsy moth larval phenology models.