Objective: To characterize informant disagreement for separation anxiety disorder (SAD). Method: The sample comprised 2,779 8- to 17-year-old twins from a community-based registry. Children and their parents completed a personal interview about the child's psychiatric history. Parents completed a personal interview about their own psychiatric history and a questionnaire about their marital relationship. Results: Informant agreement for SAD ranged between chance and extremely poor. Most cases of SAD were diagnosed by interview with only one informant. SAD diagnosed only by child interview was associated with an increased odds of father-rated oppositional defiant disorder, and vice versa. SAD diagnosed only by parent interview was predicted by the parental informant's history of antisocial personality disorder. SAD diagnosed only by paternal interview was also predicted by mother-rated marital conflict and dissatisfaction. Conclusions: Parents and children rarely agree about the presence of any level of child separation anxiety. A symptom "or-rule" mostly indexes diagnoses based on interview with only one informant, but the relative validity of such diagnoses remains unclear.