Empirical and theoretical literature on cooperative problem solving in preschool children suggests that integrating features of play into structured, experimental settings should increase the benefits of joint peer interactions and task performance. Four- and five-year-old peer dyads completed a playful, flexible, and child-driven building task or a more structured, adult-driven building task. As predicted, children in the playful condition built more complex structures, used more observational learning, and engaged in greater positive joint communication than did children in the structured condition. Condition differences carried over into a subsequent joint building task. Results suggest that cooperative problem-solving activities that allow children greater control of the task goals and interaction, similar to play contexts, can promote higher levels of cooperation and more effective learning and performance in young children. (Contains 6 tables.)