In this article, we propose a new way of understanding presidential election outcomes in red and blue states in 2000 and 2004, one that takes into account state-level variation in postmodern family patterns. Using data from the Statistical Abstract, Census, the American Community Survey, and National Vital Statistics Reports, we construct two measures of state patterns of postmodern family formation (a father-absent family scale, and a small/delayed family scale). We find that these patterns of postmodern family formation are powerful predictors of states' percentage of votes cast for the Democratic candidate in 2000 and 2004, even after controlling for differences in the composition of states' populations and for differences in states' economic characteristics (rates of economic growth, unemployment, and poverty). We suggest ways that this approach could contribute to the literature on how individual-level factors, such as demographic characteristics and moral values, shape voting behavior and electoral outcomes.