To identify generational differences in the dietary patterns of Brazilian adults born between 1934 and 1975. Design: A cross-sectional study from the baseline of the multicentre Brazilian Longitudinal Study of Adult Health (ELSA-Brasil) cohort. Year of birth was categorized into three birth generations: Traditionalists (born between 1934 and 1945); Baby Boomers (born between 1946 and 1964); and Generation X (born between 1965 and 1975). Food consumption was investigated using an FFQ. Latent class analysis (LCA) was used to identify data-driven dietary patterns. Setting: Brazil. Subjects: Individuals (n 15 069) aged 35–74 years. Results: A three-class model was generated from the LCA for each birth generation. Generation X presented higher energy intakes (kJ/kcal) from soft drinks (377·4/ 90·2) and sweets (1262·3/301·7) and lower energy intakes from fruit (1502·5/ 359·1) and vegetables (311·3/74·4) than Baby Boomers (283·7/67·8, 1047·7/250·4, 1756·0/419·7 and 365·3/87·3, respectively) and Traditionalists (186·2/44·5, 518·8/ 124·0, 1947·7/465·5 and 404·6/96·7, respectively). For Baby Boomers and Generation X, we found food patterns with similar structures: mixed pattern (22·7 and 29·7%, respectively), prudent pattern (43·5 and 34·9%, respectively) and processed pattern (33·8 and 35·4%, respectively). Among Traditionalists, we could also identify mixed (30·9%) and prudent (21·8%) patterns, and a third pattern, named restricted dietary pattern (47·3 %). Conclusions: The younger generation presented higher frequencies of consuming a pattern characterized by a low nutritional diet, compared with other generations, indicating that they may age with a greater burden of chronic diseases. It is important to develop public health interventions promoting healthy foods, focusing on the youngest generations.