The current study directly compared witnesses' susceptibility to suggestion across various structures of misleading interview questions. We examined four question structures that varied on numerous dimensions; whether they narrowed the response option to yes or no, whether they included highly specific detail about the witnessed event and whether they presumed the information being suggested to be true. Susceptibility to misinformation was measured by witnesses' responses to the initial interview questions and their responses to subsequent recognition questions. Interview questions that narrowed the response option and contained specific details and questions that encouraged broader responses but presumed certain information were found to be the most harmful. Participants were more likely to agree with the misleading suggestions contained in these question structures—and more likely to falsely report those suggested details at subsequent interview—than misleading suggestions contained in other question structures. The implications are discussed. Copyright © 2011 John Wiley & Sons, Ltd.