Objective. Many social scientists believe poor mothers are better off in middle-class than in poor neighborhoods, partly because the latter are deprived of important institutional resources. We test whether poor neighborhoods are more likely to lack one critical institutional resource, the childcare center. Methods. We use geocoded data on all licensed centers in the City of New York, address matched to Census tracts. We estimate logit models of presence of center in tract, testing for the linear and nonlinear effects of tract poverty level after controlling for residential instability, joblessness, ethnic makeup, and other demographic factors. We complement the analysis with documentary, interview, and ethnographic data on centers in one poor and one nonpoor neighborhood in the city. Results. We find (1) that the probability of presence of a childcare center does not decrease as poverty level increases; (2) the relationship depends strongly on funding source, with privately funded centers being less likely and publicly funded ones more likely to be present in poor neighborhoods; and (3) at least two factors affect why poor neighborhoods are more likely to have certain centers, the local state and the (often neglected) nonprofit infrastructure. Conclusions. The findings suggest that poor mothers are not necessarily better off in middle-class neighborhoods in this respect. The market assumptions underlying the initial hypothesis should be modified. More empirical research on the effect of the nonprofit sector on the prevalence of neighborhood institutions is needed.