Defense lawyers frequently claim that children's allegations of sexual abuse are false and are the product of coaching. As physical evidence in such cases is rare, the detection of false allegations is often dependent on the legal systems' truth-promoting mechanisms. Wigmore (1904/1974) claimed that cross-examination is the most effective of these mechanisms. This laboratory-based study investigated whether children can be coached to falsely allege a transgression and whether cross-examination promotes truthfulness from children who initially comply with coaching. One hundred and 49 kindergarten (Mage = 6 years 0 months) and Grade 3 (Mage = 8 years 10 months) students participated individually in a staged event. Participants were allocated to 1 of 3 experimental conditions. In the first condition, children were coached to allege an unwitnessed transgression, in the second condition, they were coached to deny a witnessed transgression, and in the third condition, they witnessed a transgression and were not coached. Participants then underwent a direct examination (Interview 1) followed by a repeat direct examination or a cross-examination (Interview 2). Although most children complied with coaching in Interview 1, this coaching was undermined by cross-examination in Interview 2. Consistent with Wigmore's (1904/1974) claim, cross-examination was more effective than a repeat direct examination at eliciting true transgression reports from children who initially lied in accordance with coaching. Contrary to Wigmore's (1904/1974) claim, however, cross-examination led children who were not coached to recant their initial true allegations and reduced children's accuracy for neutral events. Implications for policy and future research are discussed.