Animal models have assisted in understanding the mechanisms of brain injury underlying cerebral palsy. Nevertheless, no such models replicate every aspect of the human disease. This review summarizes the classic and more recent studies of the neuropathology of human perinatal brain injury most commonly associated with cerebral palsy, for use by researchers and clinicians alike who need to analyze published animal models with respect to their fidelity to the human disorder. The neuropathology underlying cerebral palsy includes white-matter injury, known as periventricular leukomalacia, as well as germinal matrix hemorrhage with intraventricular extension, and injury to the cortex, basal ganglia, and thalamus. Each has distinctive features while sharing some risk factors, such as prematurity and/or hypoxia-ischemia in the perinatal period. Periventricular leukomalacia consists of diffuse injury of deep cerebral white matter, with or without focal necrosis. Recent work directly in human postmortem tissue has focused on the role of free radical injury, cytokine toxicity (especially in light of the epidemiologic association of periventricular leukomalacia with maternofetal infection), and excitotoxicity in the development of periventricular leukomalacia. Premyelinating oligodendrocytes, which predominate in periventricular regions during the window of vulnerability to periventricular leukomalacia (24-34 postconceptional weeks), are the targets of free radical injury, as determined by immunocytochemical markers of lipid peroxidation and protein nitration. This maturational susceptibility can be attributed in part to a relative deficiency of superoxide dismutases in developing white matter. Microglia, which respond to cytokines and to bacterial products such as lipopolysaccharide via Toll-like receptors, are increased in periventricular leukomalacia white matter and can contribute to cellular damage. Indeed, several cytokines, including tumor necrosis factor-a and interleukins 2 and 6, as well as interferon-g, have been demonstrated in periventricular leukomalacia. Preliminary work suggests a role for glutamate receptors and glutamate transporters in periventricular leukomalacia based on expression in human developing oligodendrocytes. Germinal matrix hemorrhage, with or without intraventricular hemorrhage, occurs in premature infants and can coexist with periventricular leukomalacia. Studies in human germinal matrix tissue have focused on maturation-based vascular factors, such as morphometry and expression of molecules related to the structure of the blood-brain barrier. Gray-matter injury, seen more commonly in term infants, includes cortical infarcts and status marmoratus. Subtle cortical injury overlying periventricular leukomalacia is the subject of current interest as a possible substrate for the cognitive difficulties seen in patients with cerebral palsy. In summary, it is hoped that work in human tissue, in conjunction with experimental animal models, will lead to eventual therapeutic or preventive strategies for the perinatal brain injury underlying cerebral palsy.