The Early Head Start evaluation included 17 sites drawn from the first two waves of programs started more than a decade ago. By design, the Administration on Children, Youth and Families (ACYF) selected programs that would reflect the range of service options and context of all extant program rather than choosing a representative sample. The sites were distributed across the country. This was the first impact evaluation of services for poor pregnant women and families with children under age 3 in which the program offered center-based services in some sites and at least some home visiting in all sites. Overall, averaging across all program sites and all children and families in the sample, Early Head Start programs showed significant impacts on a wide range of child and parent outcomes when the children were 2 and 3 years old. These included impacts in cognitive, language, and social-emotional development (such as reduced aggressive behavior problems), and approaches to learning (including attention and engagement). The effects tended to appear as early as age 2 and were, for the most part, maintained through age 3. Two years later, significant impacts continued to be seen in the social-emotional (reduced behavior problems) and approaches to learning domains. However, the former Early Head Start group did not continue to show the impacts on vocabulary seen in the earlier years, except for the children who were still tested in Spanish a the prekindergarten follow-up, and Early Head Start children did not differ from control group children on measures of school-related achievement. Equally important were the impacts on parenting and the home environment, as these are crucial mediators of young children's development. The program enhanced parental support for children's language and literacy development, daily reading, and teaching activities at ages 2 and 3, with, for the most part, these effects continuing through age 5. Growth curve analyses demonstrated that the Early Head Start program had a positive impact over time in four areas. These analyses showed that for children's cognitive ability and aggressive behavior, and for maternal supportiveness and the home learning environment (1) the positive program impacts appeared early and (2) the magnitude of the impacts remained relatively constant from age 2 to age 5. As other early interventions have found, while it is noteworthy that the program impacts did not diminish with time, neither did they increase (e.g., Barnett 1995; Brooks-Gunn 2004). (Contains 3 tables. Appended are: (1) References; and (2) Tables and Figures.)