This study investigates the sustained effects into kindergarten and grade 1 of Project Head Start for disadvantaged black children. Participation in generic Head Start programs was compared to both no preschool and other preschool experience for disadvantaged children in two American cities in 1969-1970. Incorporating both pretest/posttest and comparison group information, the study has advantages over other Head Start impact studies. Both preprogram background and cognitive differences were controlled in a covariance analysis design, using dependent measures in the cognitive, verbal, and social domains. Children who attended Head Start maintained educationally substantive gains in general cognitive/analytic ability, especially when compared to children without preschool experience. These effects were not as large as those found immediately following the Head Start intervention. Findings suggest an effect of preschool rather than of Head Start per se. Initial findings of greater effectiveness of Head Start for children of below average initial ability were reduced but not reversed. The diminution of effects over time, especially for low-ability children, may reflect differences in quality of subsequent schooling or home environment.