A preschool program in rural Bangladesh was evaluated in terms of cognitive and social outcomes of children. The preschools provided a half-day program, 6 days a week, with free play, stories, and instruction in literacy and math. Four hundred children between 4.5 and 6.5 years were assessed, half in preschools and half in villages where there were no preschools. After controlling the differences in child's age, nutritional status, mother's education, and assets: preschool children performed better than the comparison children on measures of vocabulary, verbal reasoning, nonverbal reasoning, and school readiness. On some indicators of social development during play, preschool children performed better, though not on the cognitive aspects of play. They were less likely to be stunted but did not differ on most other health variables. The mean quality score from the ECERS-R was 3.5, and this correlated with the group averages of verbal and nonverbal reasoning. Results were discussed in terms of implementing a high-quality program in rural sites of developing countries, where there was a mix of play and teacher-directed instruction.