Injury surveillance provides an opportunity to determine non-intentional injury rates for those injuries which are treated in hospital accident and emergency departments. The first full calendar year of injury surveillance in north-western greater Melbourne was used to calculate incidence rates of a wide range of injury types and causes for 1989. In a population at risk estimated to number 150,604 children aged 0-14 years, the all-cause injury mortality was 10.5/100,000 per year (95% confidence interval [Cl], 5.4-15.8). There were 8207 attendances by children from the denominator area for non-intentional injury at the three hospitals participating in surveillance, representing rates of 6437/100,000 per year (95% Cl, 6258-6616) and 4406/100,000 per year (95% Cl, 4254-4558) for boys and girls respectively, while the rates for admission to hospital were 957/100,000 per year (95% Cl, 888-1025) and 649/100,000 per year (95% Cl, 590-707). The leading causes of hospital attendance were related to injuries involving sports, play equipment, bicycles and poisoning. These rates are substantially lower than those reported from other countries. The possible reasons for this include differences in health care utilization, under-ascertainment of cases, and a real difference in injury risk. Injury Surveillance Information System codes are defined for a standard set of injuries and injury causes which may be used for future comparative studies. Problems related to assessing the reliability of injury ascertainment are discussed, and the importance of integrating injury surveillance into routine hospital information systems is emphasized.