Primary care reform has triggered a flood of demonstration projects across Canada that need to be evaluated. This presents a challenge to an evaluator who is uncertain about how to convince clinical investigators to think beyond traditional research designs toward using evaluation approaches. The purpose of this article is to describe the application and benefits of using a theory-based evaluation framework for a large evaluation of four unique models of primary care delivery in Ontario, the Comparison of Models of Primary Health Care in Ontario (COMP-PC) project. Lessons learned are drawn from the authors' experience in applying the theory-based approach, including the benefits and limitations of having a common framework to facilitate model comparison.