Children’s vegetable consumption is generally below national recommendations in the UK. This study examined predictors of vegetable intake by children aged 1·5–18 years using counts and portion sizes derived from 4-d UK National Diet and Nutrition Survey food diaries. Data from 6548 children were examined using linear and logit multilevel models. Specifically, we examined whether demographic variables predicted vegetable consumption, whether environmental context influenced portion sizes of vegetables consumed and which food groups predicted the presence (or absence) of vegetables at an eating occasion (EO). Larger average daily intake of vegetables (g) was predicted by age, ethnicity, equivalised income, variety of vegetables eaten and average energy intake per d (R2 0·549). At a single EO, vegetables were consumed in larger portion sizes at home, with family members and at evening mealtimes (Conditional R2 0·308). Within EO, certain configurations of food groups such as carbohydrates and protein predicted higher odds of vegetables being present (OR 12·85, 95 % CI 9·42, 17·54), whereas foods high in fats, sugars and salt predicted a lower likelihood of vegetable presence (OR 0·03, 95 % CI 0·02, 0·04). Vegetables were rarely eaten alone without other food groups. These findings demonstrate that only one portion of vegetables was eaten per d (median) and this was consumed at a single EO, therefore falling below recommendations. Future research should investigate ways to encourage vegetable intake at times when vegetables are not regularly eaten, such as for breakfast and as snacks, whilst considering which other, potentially competing, foods are presented alongside vegetables.