A quasi-experimental, statewide intervention targeting preschool teachers' enhancement of children's language and early literacy was evaluated. Across 2 years and 20 Head Start sites, 750 teachers participated (500 target, 250 control), with 370 classrooms randomly selected to conduct pre- and posttest assessments (10 randomly selected children per class). The inability to randomize children to classrooms was addressed by examining children's performance for teachers who were control teachers in Year 1 and target teachers in Year 2. We also compared teachers with 2 years of training with teachers with 1 year of training and with control teachers. Greater gains were found for children in target classrooms than for those in control classrooms for all skills, but particularly for language skills, in Year 2, and this varied by program site. The presence of a research-based early literacy curriculum, higher levels of teacher education, and full-day versus half-day programs were significant moderators of intervention effectiveness. The challenges of implementing a statewide initiative across programs that varied in their readiness to implement a cognitively rich experience for preschool children are discussed.