It is well established that memory impairment is almost always the first aspect of cognition to become impaired in dementia of Alzheimer type (DAT). It has become apparent, however, that memory itself comprises various subcomponents. The nature of the memory deficit in DAT, in terms of the order in which these components of memory become impaired, and the underlying cause of these impairments, have become a focus of cognitive neuropsychological research. The ability to study patients with breakdown in multiple components of memory also offers the opportunity to expand our knowledge of the inter-relationship of these cognitive domains. To study these questions, I administered a battery of neuropsychological tests, comprising tests of working memory, anterograde episodic memory, autobiographical memory, remote memory and semantic memory, to 33 DAT patients designated minimal (MMSE 24-30; n=17) and mild (MMSE 17-23; n=16) and 30 age- matched controls. Firstly, a cognitive analysis was applied to the results from the famous face and famous names tests, as discussed in Chapter 3. In Chapter 4, I investigated this question and found no instances of ability to name a famous face without being able to identify the famous person. This argues strongly against the concept of naming without semantics. In Chapter 5, I establish that while autobiographical memory is impaired even in minimal DAT, executive dysfunction only becomes apparent in mild DAT. In Chapter 6, I confirmed that the major impairment in anterograde memory in DAT is due to poor encoding, rather than storage or retrieval deficits. In Chapter 7, I present data which show heterogeneity in the pattern of rCBF in DAT. The aforementioned chapters relate to the cross-sectional neuropsychological and SPECT studies of DAT. Longitudinal studies are, however, recognised to be more sound methodologically than cross-sectional studies, particularly for staging DAT. I therefore extended the study by administering the same battery of tests to 24 of the original 33 DAT patients one year later, as described in Chapter 8. In Chapter 9, I found that the deterioration in face and name identification over one year was due primarily to loss of identifying information, while face naming impairment was due primarily to a variable retrieval impairment. I found that while memory for public figures deteriorated over one year, there was no such deterioration in autobiographical memory. This further supports the assertion that remote memory may be fractionated into autobiographical memory, and memory for public figures.