The thesis addresses the relationship between populations and individuals in the philosophy of medicine. There is a long-standing tension in the fact that randomised, controlled trials in clinical populations count among the best evidence in medicine, while the goal of medical practice is to help individual patients who may differ from population averages. To what extent is evidence from populations a sufficient guide to the treatment of individuals, and how can it perhaps be improved upon? The thesis consists of five chapters that consider this problem from several complementary perspectives. The first chapter returns to the origins of population studies in medicine: The 1835 debate on medical statistics at the Académie Royale de Médecine in Paris. It argues that the existing literature has neglected core epistemological arguments on which the debate turned, and which made population studies a much more challenging methodological development than we today appreciate. The second chapter moves to a present-day version of the debate. It asks whether physicians ought more frequently to conduct so-called n-of-1 or single-case trials with individual patients. The conclusion is that while such studies are epistemologically sound, they are less useful than they may appear at first glance. The third chapter focuses on so-called molecular network reconstruction, a type of mechanistic discovery strategy that leverages large datasets. While it is often difficult to find strong associations between genetic variants and individual health outcomes, this literature suggests that a higher level of organisation -- the state of entire molecular networks -- can often be associated with individual outcomes. The fourth chapter presents an extended case study of periodontal disease, a common affliction that is well understood in some respects but also presents inter-individual heterogeneity that has been recalcitrant to explanation for decades. The puzzles of the case study lead into the fifth and final chapter, which locates the search for inter-individual variation in disease susceptibility and therapy response in an evolutionary context. It argues that evolutionary models of ultimate disease causation can serve as a heuristic tool for the study of the proximate causes of variation in health outcomes.