The present study assessed the relationship between breakfast frequency and measures of obesity in Hong Kong Chinese children aged 9–18 years. A total of 11 570 children (50 % boys) underwent anthropometric measurements and completed a simple self-administered dietary behaviour questionnaire. Their parents completed a questionnaire providing demographic information. Breakfast frequency was assessed by a single question, ‘How many days over the past week did you have breakfast?’ Children were categorised into three groups: skippers (ate breakfast 0–2 times/week); semi-skippers (ate breakfast 3–4 times/week); non-skippers (ate breakfast 5–7 times/week), to assess all associated characteristics. Of the 3644 primary and 7926 secondary school students, 8 % (8·7 % of boys and 6·9 % of girls) and 14 % (14 % of boys and 15 % of girls), respectively, were breakfast skippers. The prevalence of obesity among breakfast skippers, semi-skippers and non-skippers was, respectively, 9·8, 10·6 and 3·8 % (P < 0·001) for primary school students and 3·9, 3·1 and 2·4 % (P < 0·001) for secondary school students. The 12 % of Hong Kong children aged 9–18 years who skipped breakfast had higher BMI, BMI z-scores and percentage of body fat (PBF) than their counterparts. The dose effects of breakfast frequency (unstandardised regression coefficient, P < 0·001) on BMI and PBF were, respectively, − 0·125 kg/m2 and − 0·219 % for boys and − 0·165 kg/m2 and − 0·353 % for girls, adjusting for physical activity per additional breakfast meal per week. Further study is recommended to elucidate whether regular breakfast consumption may have a role in the prevention of childhood obesity.