This paper presents a trade-off based engineering analysis modeling (TEAM) methodology to guide design engineers towards "optimal" analysis models in a methodical and consistent manner based on their preferences. This formal approach enables quantification of abstract modeling attributes, such as cost and accuracy, for the purposes of identifying and evaluating a set of candidate models, as well as selecting and validating the choice of the most desirable model from a maximum expected utility perspective. This methodology is explored in the context of finite element modeling. Development of decision models based on cost and accuracy attributes are discussed under conditions of uncertainty. An innovative concept of quantifying the cost and accuracy of finite element models by geometric feature-based analysis is presented. Issues central to the successful implementation of a decision-based modeling procedure, such as the development of an a priori feature-based modeling knowledge base to facilitate estimation of cost and accuracy of finite element models are discussed. An example problem of  finite element modeling for structural analysis of a plastic part is used to illustrate the proposed trade- off based modeling process. Modeling results for the example problem based on two different design engineers' perspectives are presented. The results highlight the importance of treating engineering analysis modeling as a decision making process by incorporating the preferences of individual design engineers.