This paper uses the random assignment of poor families to treatment and control conditions in the Comprehensive Child Development Program (CCDP) to isolate the causal effect of center-based child care enrollment on child well-being. Operating throughout the early-1990's, the CCDP demonstration aimed to improve child development and family functioning by offering those in the treatment group five years of high-quality child care along with case management. As a result, treated children were substantially more likely to be enrolled in center-based programs throughout the preschool-age years, and I use this variation to estimate the impact of center care on children's language and social skills as well as health. I uncover mixed results: more time spent in center-based settings improves language skills but reduces social skills in the short-run, and both effects fade-out for most children within one to two years. I also find that early center care use is strongly predictive of later Head Start enrollment, indicating that a more deliberate "family retention strategy" may be effective at lengthening children's exposure to high-quality early education.