With one or two notable exceptions, tree diseases have, until recently, received little attention from epidemiological modellers. The threats to natural environments, forest and plantation crops from a range of tree pests and diseases are changing this. Scale is a particular challenge, often with multiple scales of dispersal by trade, vectors, wind and rain. Cryptic infection is another challenge, where the chances of detecting emerging infections may be low, with the result that we become aware of invading pathogens too late to intervene effectively. Mapping susceptible hosts is a third challenge; we don’t often know where all the susceptible hosts are across a threatened landscape. Sometime we look for disease in the wrong places, we intervene at the wrong scale or in the wrong place. How can we learn to do better? I propose to use examples from sudden oak death in California, citrus diseases in Florida and Texas, ash dieback and ramorum disease in the UK to explore some epidemiological principles that can be used to inform surveillance, predict spread and to compare the effectiveness of different control scenarios.