OBJECTIVES: Using a national representative sample to identify groups within the UK male and female population over 65 years who report similar patterns of diet. DESIGN: National representative dietary survey, using 4-day weighed dietary records of men and women aged over 65 years old and living in private households in Great Britain in 1994-1995. Cluster analysis was used to aggregate individuals into diet groups. SETTING: United Kingdom. SUBJECTS: 558 women and 539 men. MAIN OUTCOME MEASURES: Consumption of predefined food groups, nutrient intakes, socio-economic, demographic and behavioural characteristics. RESULTS: Three large clusters comprising 86% of the male population and three large clusters comprising 83% of the female population were identified. Among men, the most prevalent cluster was a "mixed diet" with elements from a traditional diet and some elements from a healthy diet (48% of the male population); the second was a "healthy diet" (21% of the male population); and the third was a "traditional diet high in alcohol" (17% of the male population). Among women, the most prevalent diet was a "sweet traditional diet" (33% of the female population); the second was a "healthy diet" (32% of the female population); and the last was a "mixed diet" with elements of the traditional diet and the healthy diet (18% of the female population). There were important differences in average nutrient intakes, socio-demographic and behavioural characteristics across these diet clusters. CONCLUSIONS: Cluster analysis identified three diet groups among men and three among women. These differed not only in terms of reported dietary intake, but also with respect to their nutrient content, social and behavioural profile. The groups identified could provide a useful basis for health promotion based upon the diet clusters.