When children witness or experience criminal events, the first people they go to are generally parents. Typically, no one else is privy to these conversations, and consequently little is known about their specific content. Research has shown that children can be quite accurate witnesses at times. However, they can also incorporate information from misleading and suggestive questions into their recall, and once their event memory has been changed, children may be unable to provide accurate reports. It is important then to assess parent-child discussions about crime. In the present study, 7- to 10-year-old children watched a video of a theft and talked about it with a parent immediately afterwards, and half had a second immediate interview with a researcher. All were interviewed by a different interviewer 1 week later. Results showed that: parents relied on direct and yes/no questions; children made errors of commission in response to questions; some parents asked leading or misleading questions; children incorporated all correct information from leading questions and nearly 40% of incorrect information from misleading questions; children provided additional information when interviewed by an interviewer 1 week later; and children remained relatively accurate in their descriptions but some were more accurate with parents than with an interviewer 1 week later.