Participants were 32 preschool children ages 3-5 years, 23 elementary school children ages 9-11 years, and 42 college students ages 18-50 years. The experiment was conducted across two sessions. The first session included exposure to stimulus materials, an initial interview, and immediate follow-up questions. Another set of follow-up questions was asked in a second research session, held 6-8 days after the first session. The three groups were interviewed about a video presentation they viewed. In the experimental conditions, the interviewer contradicted information about the video supplied by the participants. Participants then answered two sets of follow-up questions: one immediately following the interview and another 6-8 days later. Results indicate that participants were more likely to ignore modifications than to correct or agree with them. Adult participants were most likely to disagree with modifications. Preschoolers were most likely to incorporate modified misinformation into subsequent answers. These findings suggest that investigative interviewers should be taught that modifying witnesses' responses is a problematic interviewing technique, since it exposes interviewers to the charge that they have unduly influenced the content of a witness' statement. In addition to training interviewers to avoid modifications, witnesses should be trained to monitor the way interviewers restate their answers. Research has shown that training on "interview rules" can improve the performance of child witnesses. Children especially should be given the chance to practice noticing and correcting modifications. 7 tables, 45 references, and appended modifications used in the interviews