With the advancement of technical progress, especially with respect to magnetic resonance imaging, patchy cerebral white matter lesions (WML) are being found with increasing frequency. The (differential) diagnosis between the two main dementias of old age, (senile) dementia of the Alzheimer type ([S]DAT) and vascular dementia (VD) is made more frequently in favour of the latter, since the detection of WML leads to support a vascular origin for dementia. The present article reviews the literature concerning X-ray computed tomography (CT) and magnetic resonance imaging (MRI) in these disorders. For comparison purposes some methodological problems must be taken into account including different scoring systems for WML severity, differences in imaging techniques and in the criteria for the selection of patients and controls. A great number of studies demonstrates a strong association of frequency and severity of WML with increasing age and presence of cerebrovascular risk factors such arterial hypertension. Some studies revealed an association with neuropsychiatric deficits including gait disorders, urinary incontinence, affective lability and reduced attention and information processing speed. In CT studies, about 30% of patients with (S)DAT had WML but 36-88% in MR studies. However, only few studies controlled for the presence of cerebrovascular risk factors. More recent studies - with improved techniques - revealed a higher frequency of (slight or moderate) WML in the (S)DAT group compared to controls. The prevalence of WML in VD patients was 75-97% in CT studies and about 100% in MR studies. Therefore, without the presence of WML, the diagnosis of VD is currently in doubt. A number of in vivo investigations proved consistently - and with different methods - that cerebral blood flow was reduced in WML regions. As shown in some studies the neuropathologic correlates of WML have in common that the relative tissue water content is increased: This includes inflammation, gliosis, complete and incomplete infarctions, dilation of the perivascular (Virchow-Robin) spaces with myelin atrophy. Thus the finding of WML in watershed areas can be understood. Three case reports serve to illustrate the problems pointed out. In conclusion, the occurrence of WML is an unspecific finding which is observed in up to 50% of the elderly. Diagnostic classification as "vascular lesions" or signs of "vascular encephalopathy" or VD based on CT or MRI alone, should not be made.