Objective: To determine if Reactive Attachment Disorder (RAD) can be reliably identified in maltreated toddlers in foster care, if the two types of RAD are independent, and to estimate the prevalence of RAD in these maltreated toddlers. Methods: Clinicians treating 94 maltreated toddlers in foster care were interviewed regarding signs of attachment disorder at intake in an intervention program. Results: Using categorical and continuous measures, both types of RAD can be reliably identified in maltreated toddlers. Both continuous scores and categorical diagnoses indicated that a substantial minority of maltreated young children do exhibit signs of attachment disorders sufficient to meet criteria in DSM-IV and ICD-10. The two types were moderately convergent and at times co-occurred in the same child. Prevalence of RAD in this high-risk sample was 38-40%. Indiscriminate/disinhibited RAD was identified in children with and without an attachment figure. Within this maltreated group, toddlers whose mothers had a history of psychiatric disturbance were more likely to be diagnosed with attachment disorders. Conclusions: RAD may be reliably identified in maltreated toddlers. Emotionally withdrawn/inhibited and indiscriminate/disinhibited types of RAD are not entirely independent.