We investigated associations between 'healthy dietary pattern' scores, at ages 36, 43, 53 and 60-64 years, and body composition at age 60-64 years among participants from the MRC National Survey of Health and Development (NSHD). Principal component analyses of dietary data (food diaries) at age 60-64 years were used to calculate diet scores (healthy dietary pattern scores) at each age. Higher scores indicated healthier diets (higher consumption of fruit, vegetables and wholegrain bread). Linear regression was used to investigate associations between diet scores at each age and height-adjusted dual-energy X-ray absorptiometry-measured fat and lean mass measures at age 60-64 years. Analyses, adjusting for sex and other potential confounders (age, smoking history, physical activity and occupational class), were implemented among 692 men and women. At age 43, 53 and 60-64 years, higher diet scores were associated with lower fat mass index (FMI) and android:gynoid fat mass ratio; for example, in fully adjusted analyses, a standard deviation (sd) increase in diet score at age 60-64 years was associated with an SD difference in mean FMI of -0·18 (95 % CI: -0·25, -0·10). In conditional analyses, higher diet scores at ages 43, 53 and 60-64 years (than expected from diet scores at younger ages) were associated with lower FMI and android:gynoid fat mass ratio in fully adjusted analyses. Diet scores at age 36 years had weaker associations with the outcomes considered. No associations regarding appendicular lean mass index were robust after full adjustment. This suggests that improvements in diet through adulthood are linked to beneficial effects on adiposity in older age.