The diagnosis of diffuse axonal injury (DAI), which may be of considerable importance in forensic medicine, necessitates widespread sampling of the brain for histology. Because a limited sampling method for screening brains for axonal damage would be of value for medico-legal work, the authors have tested the findings of an earlier study which suggested that a standard set of three blocks from above and below the tentorium could reliably be used in routine practice as a basis for the diagnosis of DAI. A series of 22 previously diagnosed cases of DAI, with a range of survival times, was studied using immunohistochemistry with antibodies to beta-amyloid precursor protein (beta APP), the microglial-associated antigen CD68 (PG-M1) and for GFAP. Strict histological criteria were used to assess traumatic damage, and the evolution of the histological changes with increasing survival is described. In four cases, the sampling scheme employed yielded evidence of axonal damage in only one block, and a diagnosis of DAI could have been made in only 13/22 cases. In six of the shortest surviving cases, beta APP positivity in the corpus callosum and brainstem outlined areas of early ischaemia, as well as of traumatic damage, so that interpretation of immunolabelling was not always clearcut The findings suggest that DAI cannot be reliably diagnosed on a restricted number of blocks from vulnerable areas, and that the use of beta APP and PG-M1 immunocytochemistry may bring interpretative problems that can only be resolved by taking a larger series of tissue samples for histology.