Growing evidence supports pretense as a positive predictor of executive function (EF) in early childhood. However, there is a need for well controlled, experimental studies examining the effects of various styles of pretense on EF development in diverse populations. The present study included 179 preschoolers (ages 2–5; 38% low-income, in Head Start), randomly assigned to one of four conditions: Fantastical Pretense, Realistic Pretense, Non-Imaginative Play, or Business-as-Usual Control, with no pre-test group differences. After 5 weeks of daily intervention, data suggest that fantastical pretense, but not other styles of pretense/play, facilitates EF development among non-Head Start, middle-class children. Head Start children did not benefit, perhaps due to lower levels of engagement as well as lower initial EF levels and propensities towards pretense. These data highlight the value of engaging in fantastical pretend play among middle-class populations and emphasize the need to further investigate the role of pretense in at-risk samples. (author abstract)