How epidemiology can play a bigger role in policy-making. I am calling here for evidence-based policy, specifically policy based on epidemiologic evidence. As a discipline, we must broaden our expertise, to include a greater knowledge of policy and its formation on the one hand, and of appropriate epidemiologic methods on the other. The latter include a rehabilitation of descriptive epidemiology, better use of population health data (including administrative data), emphasis on the social determinants of health (since these are what government policies can directly influence), and health and disease modelling. We need to import several techniques from the social sciences, including geographical information systems and multilevel modelling. Demography is particularly important: since policy is implemented in real populations, the underlying population in the denominator is as important to population health as the epidemiologic events in the numerator. Borrowing from economics is already well underway, by way of economic analyses and methods for determining the utilities of various health states (needed for QALYs, etc.).