To determine the validity of using hospital-based pediatric trauma registry data to draw specific inferences with regard to regional pediatric trauma system design, we compared statistical data on the incidence and mortality of pediatric and adult injuries and burns calculated by the New York State Department of Health, based on legally mandated reports of injury deaths and hospital discharges for 1989. During this year, some 488 children, aged 0 to 14 years, died as a result of injuries, a rate of 13.8 per 100,000 annually, of whom 408 ( 11.6 100,000 ) died as a result of traumatic injuries or burns, a population-based rate 20% of that observed in adults. During the same period, 16,402 children were hospitalized for treatment of traumatic injuries and burns, a rate of 465 per 100,000 annually, a population-based rate 56% of that observed in adults; and of this number, some 90 children died, yielding an in-hospital mortality “rate” (ie, case fatality ratio) of 0.55%, and a population-based rate of 2.6 per 100,000 annually. Thus, 9.0 of the 11.6 per 100,000 children who died in New York State in 1989 as a result of traumatic injuries and burns were not admitted to the hospital and, therefore, were unknown to the statewide hospital reporting system. Detailed analysis of data by specific cause of injury was limited by the fact that hospitals in New York State did not begin reporting data on etiology (ICD-9 E-codes) as well as anatomic nature (ICD-9 N-codes) of serious injury until 1990; however, although it appears that no single injury subtype is responsible for the large number of not-in-hospital deaths statewide, homicide predominates in the metropolitan area, and motor vehicle-related injuries in others. Our results indicate that: (1) although death is a less common outcome of major trauma among children than adults, the need for in-hospital treatment of serious injuries approaches that of adults; and (2) in missing large numbers of not-in-hospital deaths (78% of the total in New York State in 1989), hospital-based data (on which most pediatric trauma registries throughout this nation are based) substantially underestimate the extent and severity of childhood traumatic injuries and burns as public health problems. We conclude that although pediatric trauma registry data are invaluable in judging treatment outcome, they should be used in guiding regional pediatric trauma system design only when considered together with population-based mortality and morbidity data derived from state or local vital statistics. Thus, collaboration with regional governmental health agencies is essential to ensure both proper data collection, and that the needs of the public for pediatric injury prevention education and pediatric trauma and burn care are optimally met.