Despite almost universal agreement that children learn causalrelationships through exploratory play, little is known about howexploratory play might support accurate causal learning. Children arepoor at designing informative experiments and there has been littleevidence for systematic patterns in children's exploratory behavior.Here I will suggest that, although the particular actions childrentake in exploratory play are quite variable, children A. are sensitiveto the availability of informative interventions and B. engage in moreexploration either when evidence fails to disambiguate equallyplausible hypotheses or when evidence provides support for apreviously implausible hypothesis. In this talk, I will presentseveral studies providing evidence for rational exploratory play inpreschoolers and toddlers.