Although perinatal complications are hypothesized to be risk factors for the development of anorexia nervosa (AN), no study to date explored this issue using a discordant sibling design. This type of design allows to explore whether the risk for obstetric complications is itself a consequence of the genetic vulnerability for AN (covariation model) or whether obstetric complications increase the risk of AN independently of (additive model), or in interaction with (interaction model), the disorder's genetic liability. The presence of perinatal complications was assessed through review of the obstetric records of 60 AN subjects, 60 unaffected sisters, and 70 healthy subjects. Unaffected sisters and healthy controls were compared in relation to perinatal characteristics and complications. There was no evidence for an elevated rate of complications in unaffected siblings of AN patients. Mothers with a positive psychiatric history tended to have more perinatal complications. Perinatal complications seem to be independent risk factors that may interact with, but are not caused by, familial risk factors for AN. In terms of prevention, a particular attention should be paid to mothers with a lifetime history of psychiatric disorders.