The implementation of child care subsidies has varied widely across countries and states, as well as over time, ranging from universal to poverty-tested eligibility. I study the implications of eligibility rules for child care subsidies in a general equilibrium, overlapping generations framework where altruistic parents invest in child skill. I allow for one- and two-parent families, and endogenize family formation with a marriage market. This explicitly incorporates single mothers, who currently parent 20% of children under 5 in the United States. Using individual-level data from the US Department of Education, I estimate how mother time, father time, and non-parental child care affect child skill for each family structure. These estimates allow me to account for the differential effect of child care subsidies on one- and two-parent families. My general equilibrium framework accounts for the effect of the subsidy on government expenditures as well as the skill distribution and, through that, on endogenous tax rates. I find that universal subsidies yield ex ante welfare gains of 5.9 percentage points, while targeting child care subsidies to one-parent families or poor families yields welfare gains of 2.4 and 2.0 percentage points, respectively. Universal subsidies more fully insure newborns against the risks they face than targeted subsidies, and do not disincentivize skill investment as happens with subsidies to the poor.