Specifically, relationships between child and interviewer characteristics and interview credibility were examined. Using the Child Abuse Interview Interaction Coding System (CAIICS, Wood, 1990), 55 videotaped interviews of high-risk sex abuse cases seen at a multidisciplinary assessment center were behaviorally coded. Findings support the interrater reliability and criterion-related validity of the CAIICS. Results also show that children were initially rated as relaxed and displayed few emotional behaviors. Thus, the assumption that a credible disclosure of abuse must necessarily include the display of emotion by the child was not supported. Second, several behavioral differences between preschool and school-aged children were identified; however, no significant gender differences were found. Third, supporting evidence was found for both age and gender effects in judgments of interview credibility, with girls and school-aged children judged as more credible. Fourth, although the interviewer did engage in so-called leading behaviors, these behaviors were not found to be related to rating of interview credibility; however, interviewer behaviors may have affected interview credibility through an intervening variable. Finally, implications for further use of the CAIICS for examining interviewer-child interactions, evaluating standards of practice, and assisting with interviewer training are discussed. 2 tables and 64 references