To determine whether living in a food swamp (≥4 corner stores within 0·40 km (0·25 miles) of home) or a food desert (generally, no supermarket or access to healthy foods) is associated with consumption of snacks/desserts or fruits/vegetables, and if neighbourhood-level socio-economic status (SES) confounds relationships.

Cross-sectional. Assessments included diet (Youth/Adolescent FFQ, skewed dietary variables normalized) and measured height/weight (BMI-for-age percentiles/Z-scores calculated). A geographic information system geocoded home addresses and mapped food deserts/food swamps. Associations examined using multiple linear regression (MLR) models adjusting for age and BMI-for-age Z-score.

Baltimore City, MD, USA.

Early adolescent girls (6th/7th grade, n 634; mean age 12·1 years; 90·7 % African American; 52·4 % overweight/obese), recruited from twenty-two urban, low-income schools.

Girls’ consumption of fruit, vegetables and snacks/desserts: 1·2, 1·7 and 3·4 servings/d, respectively. Girls’ food environment: 10·4 % food desert only, 19·1 % food swamp only, 16·1 % both food desert/swamp and 54·4 % neither food desert/swamp. Average median neighbourhood-level household income: $US 35 298. In MLR models, girls living in both food deserts/swamps consumed additional servings of snacks/desserts v. girls living in neither (β=0·13, P=0·029; 3·8 v. 3·2 servings/d). Specifically, girls living in food swamps consumed more snacks/desserts than girls who did not (β=0·16, P=0·003; 3·7 v. 3·1 servings/d), with no confounding effect of neighbourhood-level SES. No associations were identified with food deserts or consumption of fruits/vegetables.

Early adolescent girls living in food swamps consumed more snacks/desserts than girls not living in food swamps. Dietary interventions should consider the built environment/food access when addressing adolescent dietary behaviours.