The use of computer models in cotton insect management has a long history which now spans more than 30 years. In the course of this history, many people have played many roles and performed numerous research tasks. Many computer models have been written, some have been discarded, others revised, and only a few have reached the end-users, the cotton producers. This is interesting given that almost every group involved in building a model has agreed that the use of their model at the farm level was one of the intended goals. This lag can be explained by many causes. First, it has only been recently that hardware has developed to the point that computers with the required memory and speed are affordable. Secondly, many producers are uncomfortable in the use of computers. It cannot be ignored that some of this anxiety results from the fact that (once the input and output variables are learned) a "working knowledge" of the heart of the model is difficult to obtain for many extant models. This has resulted in a lack of trust and confidence for many producers. A tangible grasp of how information flows within a model and how the component parts function together is difficult to obtain. Thirdly, since many models lack robustness sometimes a model does not perform satisfactorily under new or extreme conditions. When this happens many users become discouraged and abandon the use of the model altogether. Potential users may never even start. This is unfortunate because the potential for new system insights occur when shortcomings in extant code result. Of course, many other reasons for the lack of wide acceptance of computer models can be listed. In this presentation, we discuss how several modelling techniques, concepts, and even distinct simulation models can be combined into a composite model using mathematical programming constructs. In this case, a mathematical program is a system of mathematical equations describing the production of a commodity and the consumption of resources in the context of limited capital and competing management alternatives. In cotton, the commodity is the production of seed cotton (i.e., yield) with an expected market price (i.e., capital) and losses occurring from insects (i.e., consumption of resources). To control these insects, several management tactics are available with different costs and control efficacies (i.e., the management alternatives). The purpose of the program is to select the alternative which maximizes (e.g., profit) and minimizes (e.g., cost) the management goals (e.g., to produce the most cotton at the least cost). The most important point about using mathematical programming is that several simulation models can be linked through their output, rather than having to be linked by procedural code. The output of these other models define the value assumed by many of the variables in the mathematical program. Other notable features of using mathematical programming concepts to integrate several distinct computer models are: 1) flexibility in defining system boundaries different from the system boundaries of each individual model, 2) conceptual clarity among the component parts can be quickly realized and explained, 3) management goals can be explicitly stated, 4) the alternative tactics, as well as the constraints, involved in reaching the management goals can be specifically listed and one can be quite adaptable in their selection, and 5) top-down improvement of the other simulation models used in, or related to formulating the mathematical program can rapidly occur. Some examples of these features are presented, and an example is solved using SAS Proc LP. When eventually embedded into a comprehensive and machine intelligent graphical user interface, we believe that many of the problems that limit the farm use of computer methods can be eliminated or alleviated. When speech recognition capabilities become available to support composite models by providing the ability for rapid data entry, we anticipate that the farm use of computer assisted decision making tools will rapidly proliferate. When this point is achieved, not only will progress in agricultural knowledge be achieved at the academic level, but individual producers will have the capability to perform innovative research to solve not only unique on-the-farm problems, but to be better able to cooperate with each other on a regional scale for management problems that many share. This shared use of models by the academic and farm communities will result in improved production capabilities beyond what either group could do alone.