The aim of the present study was to examine the association of pre-pubertal dietary energy density (ED) with both age and body fatness at the start of the pubertal growth spurt (age at take-off, ATO). Analyses included 219 DOrtmund Nutritional and Anthropometric Longitudinally Designed Study participants with sufficient height measurements to estimate ATO who provided 3 d weighed dietary records at baseline, i.e. 2 and 3 years before ATO (mean age 6·9 (sd 1·2) years). Mean energy intakes and amounts of foods/drinks consumed at baseline were derived from the records. ED (kJ/g) was calculated based on (1) all foods and drinks (ED_all), (2) foods and energy-containing drinks (ED_energy), (3) foods and milk as a drink, but no other beverages (ED_milk) and (4) foods only, solid or liquid (ED_food). Using multiple regression analyses, the association between the ED variables and ATO was investigated. Furthermore, Z-scores of BMI and fat mass index (FMI) at ATO were considered as outcomes to reflect body fatness at puberty onset. The results showed that ED at baseline was not associated with ATO, regardless of the ED method used. For example, mean ATO in the lowest v. highest tertile of ED_food was 9·3 (95 % CI 9·0, 9·5) v. 9·4 (95 % CI 9·1, 9·7) years, Ptrend = 0·8 (adjusted for sex, maternal age, birth weight, dietary protein, dietary fibre, baseline BMI Z-score). Similarly, ED was not independently associated with BMI or FMI Z-score at ATO (Ptrend = 0·3–0·9). In conclusion, dietary ED in childhood did not influence timing or body fatness at ATO in this cohort of healthy, free-living children.