The archaeological definition of a plague should be possible from skeletal populations, because the age profile of a population afflicted by a catastrophe will be different to that of a community exposed to a more normal mortality. The authors show how this can be done using a Bayesian statistical analysis.