Mortality from coronary heart disease in the United States has fallen 60 percent from its peak in the mid-1960s. Cardiologists and epidemiologists have debated whether this decline reflects better control of risk factors, including lifestyle interventions to reduce smoking or intake of dietary fats, or the power of medical interventions, including defibrillators and therapeutics such as statins. Attempts to resolve this debate and guide health policy have generated sophisticated data sets and techniques for modeling cardiovascular mortality. Neither effort has provided specific guidance for health policy. Historical analysis of the debate over the causes of the decline, concomitant with development of cardiovascular modeling, offers valuable policy lessons about tensions among medical and public health strategies, the changing meanings of disease prevention, and the ability of evidence-based research and models to guide health policy. Policy makers must learn to open up the "black box" of epidemiological models-and of their own decision-making processes-to produce the best evidence-informed policy.