Heterogeneity is of central importance in modelling epidemic and pandemic diseases and arises for a variety of reasons. Within-host heterogeneity arises due to differences in viral genomics, the speed and strength of individuals’ immune responses and viral shedding. Between-host heterogeneity arises from differences in contact type and rate, assortativity in mixing and individual behaviour, for instance symptomatic or diagnostic isolation. In this thesis, I start by exploring heterogeneity in viral kinetics of SARS-CoV-2 - first via fitting a hierarchical model to pooled longitudinal testing data. I then use a novel inferential method with single-time point randomised testing data from the REACT study in the UK to simultaneously infer viral kinetics and epidemic dynamics. This method is then built upon using pillar 2 community testing data in the UK, in which we also infer the incubation period distribution. Finally, we look at how heterogeneity affects inference of the reproduction number by deriving and applying a multi-type renewal equation which accounts for heterogeneity arising from discrete groups, for instance accounting for the different infectious profiles for vaccinated and unvaccinated groups. This is applied to SARS-CoV-2 in the UK, and Ebola in West Africa.