Avoidable mortality is a major public health concern but there has been little consensus among researchers on how it should be defined and reported. In this article two definitions of avoidable mortality are considered. These are used to present trends in avoidable deaths in England and Wales from 1993 to 2005, using two statistical indicators of mortality. Analysis of both definitions shows a substantial decline in causes of mortality that are considered amenable to medical intervention over this period. This article discusses which mortality indicator is most appropriate for measuring avoidable deaths, and highlights key issues in developing a National Statistics definition of avoidable mortality.