This dissertation suggests a novel approach to the application of computers for the development and use of models. Traditionally, computer assistance, as typified by computer simulation, has described models in the form of statements in the syntax of general-purpose or simulation-oriented programming languages. While such a description is an effective form of model for the purposes of simulation, it tends to inhibit a modeller's ability to define the model initially, and to modify it as new model-related insights are gained. It is the basic thesis of this research that, to be effective, computer-aided modelling must focus on both the representation function and the inquiry function of models. Representation is particularly important since it occurs in advance of inquiry, and is a primary use of modelling as a tool for decision support. A major contribution of this work is the characterization of computer-assisted modelling as consisting of three distinct processes: model building, model simulation and model validation. Entities and activities involved in these processes are identified and described, with emphasis on storage and management of information that they require or generate. To provide a basis for the development of computer aid to modelling, a formal framework is presented to give precise definition to these entities and activities. The value of the formal framework is demonstrated through a parallelism between its schemata and the schemata of relational database theory. The parallelism supports computer assistance in the construction and manipulation of model structures during model building. The nature of these activities requires storage, management and retrieval of information which is unavailable with model representations traditionally used in model simulation, but which is available with the new framework.