The object of much experimentation is to build or discover a suitable model for a given system. Unfortunately very little work has been published on what constitutes good strategies in these situations. This paper is an attempt to formulate an approach to this important problem of iteratively improving models in the area of chemical engineering kinetics. In this technique a statistical analysis is applied to the estimated parameters of a tentatively entertained theroretical model in such a way as to pinpoint its inadequacies, if they exist, so that it is possible to proceed in a logical manner to an appropriate modification of this model. This modified model is then analyzed in a similar way, and further modifications are suggested. In general the cycle is repeated as often as is necessary to reach an adequate model. This sequential method is illustrated by finding an adequate reaction model for the total catalytic oxidation of methane.