This paper examines three methodological issues concerning the measurement of semantic memory impairment in brain-damaged patients. Ten carefully selected patients with dementia of Alzheimer's type (DAT) and anomia were studied. A battery of perceptual tests and direct tests of semantic memory led to the conclusion that these patients represented a homogeneous group having a prominent deterioration of their semantic memory store without visual perceptual deficits. The first issue addressed in this patient group was whether verbal fluency impairment accurately reflected the loss of semantic memory. It was found that verbal fluency (generation of semantic category lists) was impaired due to two major constraints: deterioration of semantic memory store, and variable difficulties in semantic search. Verbal fluency, therefore, reflects semantic memory loss to some degree, but is not a direct test of semantic memory store in DAT. The second issue was whether semantic memory impairment in our patients conformed to the 'semantic storage disorder' syndrome hypothesized by Shallice (1987). It was shown that, consistent with this hypothesis, the patients demonstrated co-occurrence of consistency of errors, loss of semantic cueing, and preserved superordinate knowledge with loss of detailed knowledge of concept items. The third issue was whether semantic cueing and semantic priming are altered in a similar manner in DAT. It demonstrated that semantic cueing and semantic priming, using the same words whose concepts were degraded in semantic memory, yielded an entirely different pattern of results. Cueing and priming therefore may not be used interchangeably in the study of semantic loss after brain damage.