A series of experimentally-designed memory tests, involving different types of material presented through various sensory modalities, were administered in a standardized manner to 35 normal subjects and 10 brain-damaged patients. It was found that the common variance shared by these tests ranged from 9% to 28%, and in most cases, the correlations were not statistically significant. Furthermore, the quantitative as well as qualitative analyses revealed: (1) there was no significant difference between immediate recall and delayed recall (except in memory for constructional praxis) and (2) recognition was different from free recall. It is concluded that the measurement of memory functioning is test-specific (i.e. determined by the material used, modality of input, and mode of retrieval), and that the generalization of findings concerning memory functioning is unwarranted.