From ARIS:
The data in Exhibit 8 underscore the accuracy of conventional wisdom in the main: those who identify with one or another of the main religious groups are considerably more likely to be married than those who have no religion. Particularly the “no religion” group was far more likely to be either single, never married or single, living with a partner than any other group. Indeed, the “no religion” group shows the lowest incidence of marriage (just 19%) of all twenty-two groups. In sharp contrast, those identifying with the Assemblies of God or Evangelical/Born Again Christians show the highest proportions married, 73% and 74% respectively.
I also found it interesting that 11 percent of the “no religion” group refused to answer the question of marital status despite answering the other ones, whereas the none of the other groups were reluctant to do so. Given the way the percentages work out, I wonder if that reluctance could be related to the relatively high percentage of homosexuals who are irreligious.