Using data from an experimental evaluation in two Canadian provinces, we found that offering an earnings supplement to single mothers in place of welfare altered rates of marriage and cohabitation, but that the direction of the effects varied by province. Our findings suggest that research on the relationship between women's economic well-being and marital decisions at the national level is likely to mask important variation at the local level. After eliminating several explanations for the opposite effects in the two provinces, we propose that local labor markets and local policy contexts are potentially important mediating characteristics of marriage and cohabitation.