After addressing the post-modern argument that defining religion is impossible, bad or both, the case is made that functional definitions of religion are generally not definitions but assertions about the consequences of religion substantively defined. A substantive definition of religion is proposed. The relationship between ordinary and sociological language is discussed. A review of recent debates in the sociology of religion makes the point that our arguments rarely concern the definition of religion; they are much more often about the practical identification and measurement of the features of the social phenomenon which we want to study and those problems are not peculiar to the sociology of religion.
