It is well understood that models representing the interaction of processes in an ecological system are never completely “correct”.  For example, bounding conditions can limit the accuracy of a model and limited data may restrict how well a model can be tested.  An important question to ask is “Which assessment criteria does the model simulate effectively and which does it not?” For complex models this question can be extended to (1) “Which groups of criteria are simulated effectively at the same time with the same set of parameter values?” and (2) “Are there groups of assessment criteria that can never be simulated effectively at the same time with the same set of parameter values?”  The correct approach to answering (1) and (2) is to use Pareto optimality.  A Pareto optimal solution is the set of non-dominated solutions.  For example, for a model of stand development in a plant community undergoing competition one set of model parameters may effectively simulate the proportion of mortality that has occurred in the stand as well as the mean height of plants while another set of parameters also simulates proportion of mortality but fails to simulate mean height but does simulate mean plant weight which the first set of parameters did not.  Both sets of parameters belong to the set of non-dominated solutions.  We will illustrate the use of an evolutionary computation program that calculates the non-dominated set of solutions for a number of different ecological examples.  We will show the value of this approach in model development, testing alternative theories about ecological processes, and deciding which assessment criteria can inform about different components of an ecological model.  We will also illustrate how Pareto optimality can be used in assessing resource management decision models where there are multiple, and sometimes conflicting objectives.