An assessment of memory is relevant in many clinical contexts, but its objectives differ from one situation to another. An initial investigation should make it possible to define these objectives and to formulate hypotheses following the interview with the patient and/or his family and with the help of the overall anamnestic data. The tests proposed would aim at answering the questions according to the particular case of the patient: the supposed disorders, a suggested diagnosis, but also demographic data such as age and educational level. Consequently there is no across-the-board examination of memory and this assessment must be situated in a specific theoretical context. Memory is no longer considered as a single but as a compound function made up of different, relatively independent systems. The examination not only seeks to show the deficits which have a certain significance according to the clinical context, but also aims at bringing out preserved abilities which may contribute to the establishing of the diagnosis and on which rehabilitation and cognitive management may be based. In this article we present several methods of memory assessment with reference to the ideas of Tulving and Schacter, who propose five different memory systems: episodic memory, working memory, semantic memory, perceptive representation system and procedural memory. These assessments are made using standardized tools (tests, questionnaires) and experimental paradigms which have recently emerged in clinical neuropsychology (study of priming effects, of acquisition of skills). One of the present issues is the problem of reconciling the use of standard tools with that of the more sophisticated paradigms drawn up in the light of the most recent theoretical models. Another objective is to relate the findings of these tests to the difficulties the patients have in their everyday lives and so to achieve an ecological understanding of memory disorders.