The studies examined identified features of the perpetrators, the infant victims, and the conditions and deficiencies in systems that perform prevention efforts. The case series involved 90 children under 2 years old who were admitted to hospitals in South Wales and southwest England between 1992 and 1998; all were diagnosed with a subdural hemorrhage through neuroimaging or at a postmortem. Of the 90 children, a subset of 65 children were determined to have received their injuries from nonaccidental causes. This article reports information obtained from the children's medical histories, social histories, and reasons given by their caregivers for the injury. Although the explanations for injuries were often inconsistent with the degree and nature of the child's injuries and could not be verified, a number of categories for cases and profiles of the families emerged. A number of children had multiple injuries that included adult bites, cigarette burns, and slap marks, indicating intentional violence or abuse against the child. There was also evidence of parental stress factors and loss of control with a crying baby. Some of the babies had been exposed to child-care practices that were inappropriate for their developmental level, such as rough play or being left unsupervised or shaken as punishment. Primary prevention must focus on recognizing and meeting the needs of parents determined to be at high risk of becoming abusers, particularly when babies are born to families known to have abused their other children. Prenatal conferences must be held to identify prospective parents at risk of abusing their baby. Given the finding that many parents believe it is acceptable to shake a baby for resuscitation and that minor head injuries are to be expected in a young child, there is a clear need to educate parents about the dangers of shaking babies and very young children. A combination of prevention strategies is most likely to be effective. 24 references