The underlying reasons for recantation in children's disclosure of child sexual abuse (CSA) have been debated in recent years. In the present study, we examined the largest sample of substantiated CSA cases involving recantations to date (n = 58 cases). We specifically matched those cases to 58 nonrecanters on key variables found to predict recantation in prior research (i.e., child age, alleged parent figure perpetrator, and caregiver unsupportiveness). Bivariate analyses revealed that children were less likely to recant when they were (1) initially removed from home postdisclosure and (2) initially separated from siblings postdisclosure. Multivariate analyses revealed that children were less likely to recant when family members (other than the nonoffending caregiver) expressed belief in the children's allegations and more likely to recant when family members (other than the nonoffending caregiver) expressed disbelief in the allegations and when visitations with the alleged perpetrator were recommended at their first hearing. Results have implications for understanding the complex ways in which social processes may motivate some children to retract previous reports of sexual abuse.