In the experiment, effects of two conversation rules were tested. Children were informed that (1) "I don't know" was an acceptable answer and that (2) the interviewer would not be able to help them answer the question. A total of 114 children between 4 and 10 years of age watched a staged event and were then interviewed. The children attended an elementary school in Amsterdam, Netherlands, and participated in the experiment with parental approval. The two factors about which children were informed were systematically varied in the experiment using a 2x2 factorial design. Children were asked an open-ended question that gave them the opportunity to give information spontaneously and three misleading questions concerning the event. Results supported the hypothesis that the introduction of rules reduced suggestibility. Information given by children in response to the open-ended question was reliable and the amount of information given increased with age. When questions were misleading, the interviewer risked being misinformed by children. The number of incorrect answers was substantially reduced by explaining some rules of conversation in advance of the interview. Implications of the findings for interviewing child witnesses are examined. 32 references and 1 table