Alzheimer's disease is characterized by memory deficits. However, only a few studies have explored how patients judge their memory difficulties. The main aim of this thesis is to determine in more details whether or not Alzheimer’s patients have impaired metamnesic abilities (knowledge about own memory). Indeed, previous studies showed a complex pattern of results and suggested a fractionation of metamemory in Alzheimer’s disease (Souchay, 2007). Therefore, the main aim of this thesis is to explore this fractionation in more details and to explore whether it can be observed (1) between long-term memory tasks and short-term memory tasks, (2) according to the type of response (judgment of learning or judgment of forgetting, or (3) with implicit measures).To study the fractionation hypothesis, 7 studies have been carried out comparing younger adults, older adults and patients with amnesic Mild Cognitive Impairment or Alzheimer's disease. Results first contradicted the existence of fractionation of metamemory, Alzheimer’s patients were as accurate as predicted their performance on short-term memory tasks than on long-term memory (yet failed). Furthermore, this work showed that implicit metamemory judgments are preserved in Alzheimer's disease (Mograbi and Morris, 2013) on both episodic and semantic memory tasks, contrary of the observation on explicit measures in literature (Souchay, 2000). Finally, patients were more inaccurate than elderly when asked to predict their remembering and their forgetting on both episodic and semantic memory tasks. In conclusion, results contradict the existence of a fractionation of metamemory in Alzheimer's disease. This finding will contribute to inform metamemory (Flavell, 1979 ; Nelson et Narens, 1990) and ansognosia models (Cognitive Awareness Model : Agnew et Morris,1998 ; Hannedotir et Morris, 2007 ; Morris et Hannesdotir, 2004 ; Morris et Mograbi, 2013).