Animals are distinctive in that they are the causal agents of their own actions (e.g. a dog moves itself), whereas artifacts generally are not (e.g. a marble doesn't move itself). We examined whether children make use of this conceptual link between animacy and agency when interpreting the verb ‘move’ in English. Specifically, we hypothesized that the semantic interpretation of ‘move’ would differ, depending on whether the subject noun refers to an animal or to an inanimate object. We hypothesized that, for inanimates, children would allow ‘move’ to have a patient subject (e.g. ‘the marble moved’ could mean ‘the marble was moved by someone else’) but not so for animates (e.g. ‘the dog moved’ could not mean ‘the dog was moved by someone else’). In two studies, 65 three-year-olds, 57 five-year-olds, and 74 adults viewed video clips of animals or inanimate objects being transported by a person. For each clip, the child was asked whether the animal or object was moving. A ‘yes’ response would indicate acceptance of a patient subject (e.g. ‘the dog/marble moved’ means ‘the dog/marble was moved by someone else’). Both five-year-old children and adults more often reported that the toys were moving, than that the animals were moving. However, three-year-olds showed no animacy effects. Thus, between the ages of three and five, children begin to link animacy and agency in language. These findings suggest that children's language use is guided by similar conceptual constraints as those of adults, and/or that children are sensitive to distributional information linking form and meaning in the input language.