We evaluated the risk of brain tumor occurrence in relation to previous head injury in a population-based case-control study of 540 children with a primary brain tumor and 801 control children. The risk of a brain tumor among children with a previous head injury that resulted in medical attention was slightly elevated when compared with children with no reported head injury [odds ratio (OR) = 1.4; 95% confidence limits (CL) = 1.0, 1.9]. This effect was stronger when we restricted the head-injured group to the few children with loss of consciousness (OR = 1.6; 95% CL = 0.6, 3.9) or an overnight admission to a hospital (OR = 1.7; 95% CL 0.7, 4.6), relative to those with no head injury. We observed no appreciable association between brain tumor occurrence and birth injury involving the head or a forceps delivery. Among the few children with either a birth injury or forceps delivery and a subsequent head injury, we observed approximately twofold elevations in risk. The OR was 2.6 (95% CL = 1.1, 6.9) for those with a birth injury and subsequent head injury, relative to those with neither a birth injury nor head injury. Our results provide only weak evidence in support of head injury as an etiologic agent for brain tumor occurrence in children, although most of our exposed group had only mild head injury.