ABSTRACT Rhizomania disease of sugar beet represents a major economic threat to the sugar industry in the United Kingdom. Here we use the UK rhizomania epidemic as an exemplar of a range of highly infectious spatially heterogeneous diseases. Using a spatially explicit stochastic model, we investigated the efficacy of a spectrum of possible control strategies, both locally reactive and national in character. These include the use of novel cultivars of beet with different responses to infection, changes in cultivation practice, and reactive containment policies at the farm scale. We show that strictly local responses, including a containment policy similar to that initially implemented in the United Kingdom in response to the disease, are largely ineffective in slowing the spread because they fail to match the natural scale of the epidemic. Larger spatial-scale processes are considerably more successful. We conclude that epidemics have intrinsic temporal and spatial scales that must be matched by any control strategy if it is to be both effective and efficient. We have generated probability distributions for the proportion of farms symptomatic. Over the course of the epidemic, such distributions develop a bimodality that we hypothesize to correspond to the matching of spatial heterogeneity in the susceptible population to the intrinsic scales of the epidemic.