To determine the value of using anatomically correct dolls in diagnostic interviews of young children suspected of being sexually abused, the records of 83 children who were less than 7 years of age and who were evaluated at Yale-New Haven Hospital because of a suspicion of sexual abuse were reviewed. The dolls were used in 60 cases (72%). When the dolls were used, children provided significantly more information than by interview alone about what had happened and about the identity of the suspected perpetrator. Children less than 3 years of age, however, were unable to provide details about the abuse despite the use of the dolls. The ratings of the likelihood that sexual abuse had occurred were based on all of the information in the case including that obtained through the diagnostic interview with the dolls. When these ratings were compared with the ratings based on evidence obtained solely from noninterview data, the likelihood of abuse was higher in 35% of the cases. It was concluded that substantially more information is provided by young children when anatomically correct dolls are used and that the likelihood of detection of abuse is increased when information from the child is included in the assessment.