We used a deterministic SIR (susceptible–infected–recovered) model [27] with two age classes: children (20% of the population, a typical proportion for ages 0–18 years in a developed population) and adults (80%). The initial doubling time was set to 5 days with a 2.6 day generation time. These parameters imply a basic reproductive number of 1.4 (for this model [5]). The seed was equivalent to one infectious individual in a population of 7 million at time 0, and mixing between age groups was consistent with contact diary data for the UK (children defined as aged<20 y) [28]. The shaded regions show daily incidence of symptomatic cases for children (red) and adults (green). We assumed that 86% of infections were symptomatic [8]. The black line is the estimated number of hospital beds required at a given time. The susceptibility of children relative to adults was parameterized using the ratio of child cases to adult cases during the exponential phase of epidemic growth. (A) Baseline scenario. The ratio of early cases was proportionate to the population (20∶80, children∶adults) and all ages were equally likely to require hospitalization. (B) A scenario likely to be closer to current nH1N1 dynamics. The ratio of early cases was 50∶50 and adults were much more likely to require hospitalization.