Objective: Mental health professionals frequently classify children's attachment style using a combination of the parent/caregiver interview, an interview with the child, the interviewer's clinical impressions, and at times, the child's responses to projective tests not aimed at eliciting attachment specific information. However, no studies have been conducted on the validity or reliability of attachment classifications based on developmental histories. This study aimed to investigate the accuracy of history-based attachment classifications. Method: Attachment classifications based on intake interviews, administered to 35 eight to twelve-year- old children and their caregivers attending an outpatient mental clinic service in a government hospital in Johannesburg, were compared to classifications obtained through the Attachment Story Completion Test (ASCT). The level of agreement between rates and across assessment measures was analysed. Results: The association between ASCT and history-based attachment classifications was poor. Conclusion: The possible misclassification of children's attachment styles according to histories has implications for both diagnosis and treatment.