The level of under-reporting of injury events which resulted in medical treatment was determined for a group of 631 thirteen-year olds. Two-year recall data was compared with the medical records of an accident and emergency department. Thirty-nine percent of all visits to the accident and emergency department were not recalled. Recall was shown to be related to the number of injury events, time elapsed since an injury event, type and severity of injury but not to whether the victim was hospitalized. Logistic regression analysis suggested that the recall of injury events depended on the type of injury sustained, time elapsed since the injury event, the number of injury events, and an interaction effect between the time and the number of injuries.