The current study of preschool children characterizes a semi-natural extension of experimental questions on how human ancestors evaded predation when encountering dangerous felids. In a pretend game on a playground, we presented full-size leopard and deer models to children (N = 39) in a repeatedmeasures experimental design. Prior to viewing the model presented 15-m away, each child was instructed by the experimenter to go where she or he would feel safe. The rationale for this study was based on the anthropological construct of "sexual dinichism," positing that, during the Pliocene, smaller-bodied hominin females engaged in more arboreal behavior than larger-bodied males. Consistent with this construct, our previous simulation research using images of an African rock outcrop showed that, after viewing a lion, girls preferred a tree as refuge rather than a crevice or large boulder whereas boys did not differentiate these refuge sites. In this follow-up study, we predicted that, after viewing the model leopard, the preschool girls would differ from the boys by not choosing enclosed refuge sites analogous to the crevice. Analyses of a contingency table for the leopard model supported this hypothesis by yielding a significant interaction of sex and refuge location (p = .031, d = .76), the source of which was a reliably larger percentage of girls not choosing concealed refuge (p = .005, d = 2.3). The interaction of sex and refuge location for the model deer was not significant (p > .5). Our findings suggest that, in contrast to the deer, the girls selected exposed playground refuge sites rather than concealing ones to maintain visual contact with the leopard as a contingency for future action.