Institutions such as hospitals and nursing or long-stay residential homes accommodate individuals at considerable risk of mortality should they acquire SARS-CoV-2 infection. In these settings, polymerase chain reaction tests play a central role in infection prevention and control. Here, we argue that both false negative and false positive tests are possible and that careful consideration of the prior probability of infection and of test characteristics are needed to prevent harm. We outline evidence suggesting that regular systematic testing of asymptomatic and pre-symptomatic individuals could play an important role in reducing transmission of SARS-CoV-2 within institutions. We discuss how such a programme might be organised, arguing that frequent testing and rapid reporting of results are particularly important. We highlight studies demonstrating that polymerase chain reaction testing of pooled samples can be undertaken with acceptable loss of sensitivity, and advocate such an approach where test capacity is limited. We provide an approach to calculating the most efficient pool size. Given the current limitations of tests for SARS-CoV-2 infection, physical distancing and meticulous infection prevention and control will remain essential in institutions caring for vulnerable people.