The settings for the study were a tertiary care children's hospital and county medical examiner's office. The patients were 12 children with intracranial injury secondary to child abuse. None was clinically suspected to have cord injury. The subjects included all eligible children whose attending physicians believed they needed follow-up cranial imaging, could be safely imaged, and whose caretakers consented between November 1991 and September 1994. Interventions involved MRI scans of the cervical spine either more than 3 days after clinical presentation or postmortem. The main outcome measures were clinical observations by a neurologist, child protection team pediatrician, and medical examiner by prospective protocol. MRI scans were evaluated by prospective radiology protocol with an emphasis on cervical cord injury. Four of the five autopsied children had small subdural or subarachnoid hemorrhages at the level of the cervical spine; MRI scan did not identify them. Neither did MRI identify cord injury in any child studied. The study concluded that routine cervical spine MRI scans are probably not warranted in children with head injury secondary to child abuse without clinical symptoms of cervical cord injury. 1 table and 16 references