During the seven-year period from 1976 through 1982, 2133 cervical spine radiographs were obtained for children less than 18 years of age at two hospitals in Tucson. Twenty-five children (1.2%) had x-ray evidence of a cervical spine injury, with a male to female ratio of 4:1. Vehicular accidents accounted for 36% of cervical spine radiographic abnormalities, and sports or playground accidents accounted for an additional 36%. In an attempt to find clinical predictors that might identify x-ray abnormalities, we reviewed the medical records of these 25 children with abnormalities and 713 randomly selected children without x-ray evidence of cervical spine injuries for the following: method of injury, presenting complaints, physical examination findings, therapy, and complications. No single clinical predictor had a sensitivity of 100% when considered in isolation, but clinical assessment consisting of EITHER a complaint of neck pain OR involvement in a vehicular accident with head trauma would have correctly identified all 25 cases of cervical spine injury. If this information had been used prospectively, the number of cervical spine radiographs ordered would have been reduced by 32%. We conclude that the use of this clinical "marker" would have positively identified all children with cervical spine injuries and would have reduced by one third the cost and radiation exposure associated with cervical spine radiographs. Because of the serious consequences of missing a cervical spine injury, we suggest that other studies confirm these results before this information is accepted as a recommendation.