The purpose of the study was to determine the impact of individual and community level risks on school outcomes of children who attend Head Start. We studied a sample of 3,693 African American and Hispanic children who had been born in New York City, participated in Head Start, and attended New York City public schools. The outcome was the score obtained on a citywide third-grade reading test. Individual level risk factors were derived from birth certificate data. Community level risks were extracted from citywide U.S. Census data and other public-access data sets. Multilevel regression analyses indicated that at the individual level, lower reading scores were significantly associated with: male gender, low birth weight, unmarried mother, low maternal education, and inadequate interpregnancy spacing. Controlling for individual-level risk, concentrated community poverty significantly lowered reading scores, and a high percentage of immigrants in the community significantly raised scores. There was also a significant crosslevel effect: boys benefited more than girls from the immigrant community effect. The evidence suggests that we can better identify children at future educational risk and maximize the success of early intervention programs by exploring influences on school success at multiple levels, including the community.