Recognizing the existence of memory disorders in elderly subjects nece
ssarily involves developing a diagnostic strategy to identify the caus
es. Three types of complementary investigation are available: neurophy
siological and neurobiological tests, and anatomical and functional br
ain imaging studies. Major efforts have been made to identify diagnost
ic markers on the electroencephalogram. Quantified EEG may have a cert
ain value. However, it provides a plethora of data and there is no con
sensus on which items are most relevant on the EEG, whether the patien
t is awake or asleep. Evoked 'cognitive' potentials may also provide u
seful data, particularly to distinguish between the different types of
degenerative dementia. Exaggerated pupil dilation in response to a my
driatric drug has also been put forward as a diagnostic test. The resu
lts, however, are controversial. For the time being there is no diagno
stic laboratory parameter that can be used routinely, even if studies
of the proteins tau and P97 are promising. Molecular genetics-based st
udies have identified a number of chromosomal abnormalities in certain
families. Apart from studies of chromosome 19 and apolipoprotein 3, t
hese markers have no practical utility. Allele Sigma 4 is associated d
th a higher risk of Alzheimer's disease. Apolipoprotein E phenotyping
can of course be of diagnostic value, but no more so than neuropsychol
ogical or neuroimaging methods. On the other hand, it is difficult to
derive a predictive test It also seems difficult to develop a diagnost
ic strategy without including brain imaging studies. The value of morp
hological imaging (CT, MRI) in the diagnosis of dementias is clear. Vo
lumetric measurements of certain brain structures might be a useful di
agnostic approach for early detection. Functional brain imaging method
s (PET and SPECT) appear particularly suited to the diagnosis of degen
erative dementias. The presence of functional abnormalities at onset,
and even before the first clinical signs appear, is clearly valuable.