Effective communication entails the strategic presentation of evidence; good communicators present representative evidence to their listeners-evidence that is both consistent with the concept being communicated and also unlikely to support another concept a listener might consider. The present study examined whether preschool-age children effectively select evidence to manipulate others' semantic knowledge, by testing how children choose evidence in a teaching or deception task. Results indicate that preschoolers indeed effectively select evidence to meet specific communicative goals. When asked to teach others, children selected evidence that effectively spanned the concept of interest and avoided overly restrictive evidence; when asked to deceive others into believing a narrower concept, they selected evidence consistent with the overly restricted belief. Thus, results support the idea that preschool children possess remarkable abilities to select the best evidence to manipulate what others believe.