Résumé : Objective: To provide a consensus-based minimum set of criteria for the diagnosis of malnutrition to be applied independent of clinical setting and aetiology, and to unify international terminology. Method: The European Society of Clinical Nutrition and Metabolism (ESPEN) appointed a group of clinical scientists to perform a modified Delphi process, encompassing e-mail communications, face-to-face meetings, in group questionnaires and ballots, as well as a ballot for the ESPEN membership. Result: First, ESPEN recommends that subjects at risk of malnutrition are identified by validated screening tools, and should be assessed and treated accordingly. Risk of malnutrition should have its own ICD Code. Second, a unanimous consensus was reached to advocate two options for the diagnosis of malnutrition. Option one requires body mass index (BMI, kg/m2) <18.5 to define malnutrition. Option two requires the combined finding of unintentional weight loss (mandatory) and at least one of either reduced BMI or a low fat free mass index (FFMI). Weight loss could be either >10% of habitual weight indefinite of time, or >5% over 3 months. Reduced BMI is <20 or <22kg/m2 in subjects younger and older than 70 years, respectively. Low FFMI is <15 and <17kg/m2 in females and males, respectively. About 12% of ESPEN members participated in a ballot; >75% agreed; i.e. indicated ≥7 on a 10-graded scale of acceptance, to this definition. Conclusion: In individuals identified by screening as at risk of malnutrition, the diagnosis of malnutrition should be based on either a low BMI (<18.5kg/m2), or on the combined finding of weight loss together with either reduced BMI (age-specific) or a low FFMI using sex-specific cut-offs.