To compare the intake of key indicator foods at age 12 months and 18 months between infants of Pakistani and White British origin.

Logistic regression was used to model associations between ethnicity and consumption of key indicator foods defined by high or low energy density using an FFQ at age 12 and 18 months.

Born in Bradford 1000 study, Bradford, UK.

Infants (n 1259; 38 % White British, 49 % Pakistani), mean age 12·7 (sd 1·0) months and toddlers (n 1257; 37 % White British, 49 % Pakistani), mean age 18·7 (sd1·0) months.

At 12 months, Pakistani infants consumed more commercial sweet baby meals than White British infants, with greater odds for being above average consumers (adjusted OR (AOR)=1·90; 95 % CI 1·40, 2·56), more chips/roast potatoes (AOR=2·75; 95 % CI 2·09, 3·62), less processed meat products (AOR=0·11; 95 % CI 0·08, 0·15), more fruit (AOR=2·20; 95 % CI 1·70, 2·85) and more sugar-sweetened drinks (AOR=1·68; 95 % CI 1·29, 2·18). At 18 months these differences persisted, with Pakistani infants consuming more commercial sweet baby meals (AOR=4·57; 95 % CI 2·49, 8·39), more chips/roast potato shapes (AOR=2·26; 95 % CI 1·50, 3·43), more fruit (AOR=1·40; 95 % CI 1·08, 1·81), more sugar-sweetened drinks (AOR=2·03; 95 % CI 1·53, 2·70), more pure fruit juice (AOR=1·82; 95 % CI 1·40, 2·35), more water (AOR=3·24; 95 % CI 2·46, 4·25) and less processed meat (AOR=0·10; 95 % CI 0·06, 0·15) than White British infants.

Dietary intake during infancy and the early toddlerhood period is associated with ethnicity, suggesting the importance of early and culturally adapted interventions aimed at establishing healthy eating behaviours.