Objective. It is widely acknowledged that food intake declines with ageing, which predisposes the elderly to unintentional weight loss and nutritional risk. The present experiment intended to explore whether this decline in food intake varies depending on food category.Method. We weighted all the foods and drinks consumed by 68 institutionalized elderly people (66-100 yo; 75% of women) over 24 hours. Data were submitted to a Hierarchical Cluster Analysis.Results. We observed 4 eating patterns. A first cluster (18% of the panel) includes people for who food intake is close to recommendation. They are the largest eaters of meat (on average, they consumed 130g of meat at lunch) and fulfil 95% of the Recommended Protein Allowance (RPA). A second cluster (28%) includes people who eat very little whatever the type of food. They are the smallest eaters of meat (45g at lunch) and fulfil 60% of the RPA. A third cluster (22%) includes people who eat little meat (74g at lunch) and very little cheese but remain good eaters of soup and dairy products. Finally, a last cluster (32%) includes people who eat little meat and very little desserts, but remain good eaters of vegetable and fruit.Conclusion. Results provide insights to develop a food supply enabling the elderly to fulfil their nutritional needs while taking into account their eating pattern, i.e. the portion that they are able / willing to ingest for each food category.