Policy is formulated on the basis of values, ideology, political pressures (in many languages, the words for “politics” and “policy” are the same) and evidence. Epidemiology contributes to the evidence. Although many of us believe that policy should be largely or entirely evidence-based, there is widespread agreement that evidence (at least, scientific evidence) plays a relatively minor role. Indeed, making policy is often likened to making sausage: a process that you might not wish to observe too closely, because you might feel better if you do not know what goes into it! In view of these considerations, it is easy to become discouraged and to abandon policy in favour of some activity that promises greater use of one’s results. But this is a defeatist approach: many of the avenues to improve health can only be achieved through policy approaches, so to abandon policy is to abandon population health. Much better to provide relevant epidemiologic evidence and do what we can to ensure that as much as possible of it is used, in the belief that policy that is somewhat evidence-based is better than policy that is not at all evidence-based.