Aim of the present review paper was to evaluate the hypothesis (included in the proposal of new research criteria for Alzheimer's disease; Dubois et al., Lancet Neurology, 6, 734-746, 2007) that a neuropsychological tool which provides support for the semantic encoding of memorandum at the time of study and supplies category cues at the time of retrieval (i.e. the Grober-Buschke paradigm) is more effective than traditional measures of free recall in 1) differentiating patients affected by the amnestic form of Mild Cognitive Impairment (MCI) or by mild to moderate forms of Alzheimer's disease (AD) from healthy matches, 2) predicting the conversion of individuals with MCI to AD, and 3) differentiating AD patients from individuals affected by other forms of dementia. Results of the review are controversial regarding the superiority of the Grober-Buschke procedure in differentiating individuals affected by AD or MCI from healthy individuals. The only study that evaluated this issue directly found that the Grober-Buschke procedure was more sensitive and specific than more traditional memory tests in predicting the conversion of MCI patients to AD. Finally, two studies reported that patients affected by AD or other forms of dementia showed different performance patterns in the free and cued recall tasks of the Grober-Buschke procedure. In conclusion, although encouraging results are reported in the few studies that investigated the ability of this procedure to predict the evolution of individuals with amnestic MCI and to differentiate AD patients from patients with other forms of cortical and subcortical dementia, more experimental work is needed to confirm these positive findings.