Background: Dietary patterns provide a summary of dietary intake, but to our knowledge, few studies have assessed trends in dietary patterns in the population.Objective: The aim was to assess 20-y trends in dietary patterns in a representative sample of the Geneva, Switzerland, population with the consideration of age, sex, education, and generation.Design: Repeated, independent cross-sectional studies were conducted between 1993 and 2014. Dietary intake was assessed by using a validated food-frequency questionnaire. Dietary patterns were assessed by using principal components analyses.Results: Among 18,763 adults, 1 healthy ("fish and vegetables") and 2 unhealthy ("meat and chips" and "chocolate and sweets") patterns were identified. Scores for the "fish and vegetables" pattern increased, whereas the "meat and chips" and "chocolate and sweets" pattern scores decreased in both sexes and across all age groups. The stronger increase in the "fish and vegetables" pattern score among the less well-educated participants led to a narrowing of educational differences (mean ± SD scores in 1993: -0.56 ± 1.39 compared with -0.05 ± 1.58 in low- compared with highly educated groups, respectively; P < 0.001; scores in 2014: 0.28 ± 1.64 compared with 0.24 ± 1.83, respectively; P = 0.772). Generational analysis showed that older age groups tended to show smaller changes than younger age groups: the yearly score change in "chocolate and sweets" was -0.021 (95% CI: -0.027, -0.014; P < 0.001) for the 35- to 44-y cohort compared with -0.002 (95% CI: -0.009, 0.005; P = 0.546) for the 45- to 54-y cohort.Conclusions: Three dietary patterns were identified; scores for the "fish and vegetables" pattern increased, whereas the "meat and chips" and the "chocolate and sweets" pattern scores decreased. The stronger increases in the "fish and vegetables" pattern score among the less well-educated participants led to a smaller difference in dietary intake across the different educational levels.