Previous modeling efforts using dynamic intervention modeling have relied on a priori postulations of the model form, thus perhaps incorporating an experimentor's subjective biases. An algorithm for the identification of a tentative model formed from the data alone is reviewed. Previous intervention analyses not using an identification phase are revisited to illustrate the differences in resulting models and associated policy evaluations. The first example deals with controls applied to the Los Angeles County area circa 1960-66 to help alleviate a serious pollution problem evidenced by rising levels of ozone. The second illustration is directed to the evaluation of changes of property or life loss associated with motor vehicle crashes after implementation of a traffic safety program. Figures, formulas, and seven references are included. (Author summary modified)