Human cytomegalovirus (HCMV), like all herpesviruses, persists in the host after primary infection, with reactivation often occurring as a result of immunosuppression. The lack of a model system has made analysis of HCMV latency and reactivation difficult. However, the ability to analyse which specific cell types in vivo carry the virus and the relative levels of permissiveness of these cell types for viral gene expression and productive infection is beginning to help us understand these complex molecular events.