Children who had been observed in preschool when 3 years old, using an observation schedule (Fagot & Patterson, 1969) consisting of 28 play behaviors, were assigned interest scores on the basis of percent of time spent in the various activities. Masculinity scores were computed on the basis of percentage of significantly preferred sex-typed behaviors. When the children were 6 and 10 years old, they were given the Children’s Embedded-figures Test and rated by their own teachers on intellectual performance. The relationship of preschool interest patterns to elementary school academic achievement and projected career choices suggested that the play choices in preschool have different meanings for boys and girls and therefore different consequences for later achievement.