In three experiments, children?s reliance on other people?s testimony as compared to their own, first-hand experience was assessed in the domain of ontology. Children ranging from 4-8 years were asked to judge the ontological status of four different types of entity: real entities (e.g., giraffes) that correspond to everyday, natural kinds; scientific entities (e.g., germs) that are generally assumed to exist but cannot be identified without the help of specialized instruments; extraordinary beings (e.g., angels) that are believed to exist by some people but not by others; and impossible entities (e.g., flying pigs) that nobody believes in. Preliminary results showed that irrespective of age, children judged that everyone, themselves included, believes in the existence of scientific as well as real entities. Nevertheless, children justified their existence claims differently for the two types of entity and said that although they would recognize a giraffe for example, they would not recognize a germ. Children also denied the existence of impossible entities, and especially at 8 years, acknowledged the lack of consensus regarding extraordinary entities. Taken together, the findings show that children?s ontological claims are not simply based on their first-hand encounters with instances of a given category but are guided by the testimony supplied by other people. Moreover, children can form categories in the absence of critical information for knowing the extension of that category. Some of the results of this empirical research have presented at the Biennial Meeting of the Society for Research in Child Development (SRCD), Tampa, USA (Harris & Pons, 2003). This empirical research is still running. With Paul Harris (Harvard University) and Jessica Ascher (University of Leiden).