To describe dietary patterns by applying cluster analysis and to describe the cluster memberships of European children over time and their association with body composition changes.

The analyses included k-means clustering based on the similarities between the relative frequencies of consumption of forty-three food items and regression models were fitted to assess the association between dietary patterns and body composition changes.

Primary schools and pre-schools of selected regions in Italy, Estonia, Cyprus, Belgium, Sweden, Hungary, Germany and Spain.

Participants (n 8341) in the baseline (2–9 years old) and follow-up (4–11 years old) surveys of the IDEFICS (Identification and prevention of Dietary- and lifestyle-induced health EFfects In Children and infantS) study.

Three persistent clusters were obtained at baseline and follow-up. Children consistently allocated to the ‘processed’ cluster presented increased BMI (β=0·050; 95 % CI 0·006, 0·093), increased waist circumference (β=0·071; 95 % CI 0·001, 0·141) and increased fat mass gain (β=0·052; 95 % CI 0·014, 0·090) over time v. children allocated to the ‘healthy’ cluster. Being in the ‘processed’–‘sweet’ cluster combination was also linked to increased BMI (β=0·079; 95 % CI 0·015, 0·143), increased waist circumference (β=0·172; 95 % CI 0·069, 0·275) and increased fat mass gain (β=0·076; 95 % CI 0·019, 0·133) over time v. the ‘healthy’ cluster.

Children consistently showing a processed dietary pattern or changing from a processed pattern to a sweet pattern presented the most unfavourable changes in fat mass and abdominal fat. These findings support the need to promote overall healthy dietary habits in obesity prevention and health promotion programmes targeting children.