s / International Journal of Infectious Diseases 79(S1) (2019) 1-150 125 account vaccine efficacy. A range of vaccine coverages were generated using an algorithm controlled by two parameters that specify the size and depth of heterogeneities. Transmission dynamics were performed on a highly resolved spatial description of the human geographical distribution at a resolution of 5 km which included individual mobility.Results: We found that, assuming an average vaccine coverage of 92%, corresponding to the seroprevalence measured in the United States in 2015, the number of annual cases and outbreak sizes reported in recent years is compatible with the number of estimated importations combined with a highly subcritical reproductive number and a moderately heterogeneous vaccine coverage. Under these conditions, outbreaks of the size of the 2015 California outbreak are not rare. We also show that localized drops in vaccine coverage may lead to a catastrophic failure of herd immunity if they exceed a heterogeneity dependent threshold.Conclusion: Given the current average level of vaccine coverage, outbreaks with sizes similar to the 2015 California one may occur in the future. However, the fact that the effective reproductive number is highly subcritical confirms the high level of control over measles currently enforced in the United States.Purpose: Epidemics pose a significant risk for human and economic losses, yet this risk is challenging to quantify. Current methods for assessing epidemic risk are typically based on observational data, which are often sparse and incomplete. Global simulation models of infectious disease spread allow us to simulate spatiotemporal dynamics of hypothetical epidemic scenarios. Sovereign nations in Africa are at risk of experiencing filovirus epidemics, which can have devastating financial impacts. These models can help fill in the gaps in observational data and provide a clearer picture of epidemic risk. Methods & Materials:We used a stochastic epidemic model to simulate over 200,000 filovirus epidemics, with varying initial parameters and differing availability and efficacy of intervention measures. Probability distributions of model parameters such as spark location, transmissibility, and case-fatality ratio were derived from the primary literature and historic case data. We constructed an event catalog by selecting simulated events using a sampling algorithm that accounted for the joint probability of each parameter combination for each scenario and the inter-arrival time distribution between filovirus epidemics. The event catalog was then used to generate exceedance probability metrics, which estimate the likelihood of observing an event of a given severity (e.g., expected deaths), or worse, in any given year.Results: Based on our analysis, there is a 1.5% probability that a filovirus epidemic similar to or worse than the 2014 West Africa Ebola epidemic (based on global deaths) will occur in any given year. The countries that are predicted to most commonly experience filovir...