Recent studies have shown an inverse association between the level of dietary Ca, particularly from dairy sources, and body weight in adults; there is, however, a paucity of data regarding this relationship in children. We therefore investigated this issue in 1087 children who underwent body weight and height measurement during a survey on childhood obesity. Lifestyle and dietary habits were investigated by a questionnaire. After excluding children who were following a dietary regimen for any reason, 884 children (M:F 451:433; age 7.5 (sd 2.1) years) were selected. Milk consumption was pooled into four frequency categories: poor (< or =1/week; n 125), moderate (>1 but < or =5-6/week; n 133), regular (1/d; n 408) and high (> or =2/d; n 218). The frequency of consumption of milk was inversely and significantly associated (t=-2.964, P=0.003) with age- and sex-specific BMI z-scores by linear regression analysis, controlling for sex, age, physical activity, birth weight and parental overweight and education. The statistical association remained significant (t=-2.831, P=0.005) after the inclusion of children consuming only skimmed milk (n 91). Milk consumption was still significantly and inversely associated with BMI z score (t=-2.791, P=0.005) in the whole-milk consumers when controlling for age and the frequency of consumption of various foods; this association was no longer significant (P=0.21) when children consuming skimmed milk were included in the analysis. This is the first report showing a significant inverse association between frequency of milk consumption and body mass in children. Regardless of the mechanisms involved, our results might encourage further research on this issue and might have important implications for the prevention of obesity.