We compared 1- and 2-year-old children's performance on Pretend and Reality tasks. Pretend tasks involved the comprehension of a pretend scenario, whereas Reality tasks did not. For example, the experimenter pretends to drink water from an empty cup, she fills another cup with imaginary water and then invites the child to drink. In the Reality version, the experimenter uses real water in making exactly the same actions and the same request to the child. Our aim was to verify when very young children understand pretense, and to determine whether failures to understand pretense are the result of difficulties specific to pretense or not. Results showed that starting from 16 months, children begin to understand pretense. At no time did performance differ between Pretend and Reality tasks, suggesting that young children's difficulties with pretense may not arise from causes specific to pretense.