Many of us who propose and test multiparameter simulation models have developed a good deal of expertise about the process. Some of this skill has to do with highly technical matters that are well represented in the literature (such as parameter estimation programs). We have found that these sorts of well researched technical components account for a tiny fraction of the time and effort involved in model development. The great proportion of effort (both on the part of the theorist and probably also in terms of computation time) involves skills that are highly abstract and subjective, and are seldom discussed in the literature. Thus the most apt title for this talk is the art of model development. The discussion will touch upon some associated lessons for the role of computation.