Battered children represent only a small proportion of of abused children, and many deaths from child abuse occur without evidence of significant battering. What physicians and forensic pathologists see instead is a spectrum of fatal injuries whose cause often cannot be determined by autopsy or laboratory studies alone. These subtle forms of abuse are most common in infants who are especially vulnerable because of their relative isolation, small size, lack of verbal skills, and total dependence on caretakers. The pathology of child abuse is examined in relation to the autopsy (radiological documentation, time of death determination, trace evidence, nutrition and hygiene, cutaneous manifestations of abuse, bite mark injuries, thoracic injuries, abdominal injuries, musculoskeletal injuries, and sexual abuse injuries). The value of laboratory studies in the areas of post-mortem chemistry, intoxication and poisoning, and documentation of infection is also considered. Special problems associated with conducting child autopsies are noted as they pertain to fatal head injuries, the timing of head injuries, fatal abuse without fatal injury, dehydration and failure to thrive, mistaken diagnosis of child abuse, and perinatal deaths. Recognizing the autopsy alone does not always establish the cause and manner of death of a child, the investigation of child fatalities by death review teams is discussed. 105 references, 3 tables, and 16 photographs