Lockdowns, or modern quarantines, involve the use of novel restrictive non-pharmaceutical interventions (NPIs) to suppress the transmission of COVID-19. In this paper, I aim to critically analyze the emerging history and philosophy of lockdowns, with an emphasis on the communication of health evidence and risk for informing policy decisions. I draw a distinction between evidence-based and modeling-based decision-making. I argue that using the normative framework of evidence-based medicine would have recommended against the use of lockdowns. I first review the World Health Organization’s evidence-based pandemic preparedness plans for respiratory viruses. I then provide a very brief history of COVID-19 modeling, which was cited as justification for the use of lockdowns in the U.K., the U.S., and much of the world. I focus on the so-called Imperial College model designed by Neil Ferguson et al. as well as the so-called Oxford model designed by José Lourenço et al. I analyze the evidence-based pandemic response known as ‘mitigation’, and I compare it with Ferguson et al.’s experimental strategy known as ‘suppression’. I summarize the strengths and weaknesses of these strategies based on their diametric aims and each model’s parametric assumptions. Based on my critical analysis of the suppression strategy, I attempt to expose what has been called the ‘logic of lockdowns’, which Sunetra Gupta of the Oxford model group has suggested is flawed. Finally, I consider Trisha Greenhalgh’s objection to evidence-based policy based on the precautionary principle, and I attempt to offer a response. I conclude with a brief narrative review of the emerging randomized evidence on restrictive NPIs, which seems to support my claim that mitigation was the strategy that would have been recommended by evidence-based medicine. If this is true, then COVID-19 modeling may serve as an important reminder of the enduring lesson of evidence-based medicine: that one should always ‘Trust the Evidence!’ for better health policy.