Durkheims Suicide has been criticised on theoretical, methodological and empirical grounds. While some argued that Durkheim made macro- instead of micro-propositions, others maintained that he formulated micro-propositions but failed to test them with micro-data. Pescosolido (1990) proposed an alternative hypothesis that was supported empirically. In this article, we show that this hypothesis can be deduced from a general formulation of Durkheims theory. Building on Durkheims work, we formulate multilevel propositions about the impact of denomination and religious composition of municipalities on suicide in the Netherlands from 1936 until 1973. Using individual level data on suicides, we find that for every denomination in a municipality the likelihood of committing suicide decreases with the proportion of religious persons in that municipality. Furthermore, in accordance with the decline of denominational segregation in Dutch society from the 1950's, we find that the impact of religious composition of the municipality on the likelihood of committing suicide is declining. These results are in accordance with Durkheims general theory of suicide.