ABSTRACT Background : Adiponectin has been shown to be inversely related to birth weight in schoolchildren and adolescents. However, the available information regarding a relation between birth size and adiponectin in infants and preschoolers is limited. Methods : A longitudinal study was conducted among healthy Japanese children, and serum adiponectin levels were measured at consecutive visits. The effect of sex, gestational age, birth length, birth weight, or placental weight on adiponectin was analyzed by using a linear mixed model for repeated measures. Results : Eighty-three children were evaluated. The age range at the first visit was 0.07–5.3 years. A total number of 227 measurements were made, 1–12 per subject (median, 3) and the duration of follow-up ranged from 1 day to 2.3 years (median, 1.0 years). The subjects were divided into four groups according to the age at the first visit: (1) 0–2 months (n = 29), (2) 3 months to 1 year (n = 17), (3) 2–3 years (n = 19) and (4) 4–5 years (n = 18). In the multivariate model, birth length (coefficient = 3.94, 95% CI, 0.23–7.65) was an independent predictor of serum adiponectin levels in subjects aged 0–2 months. Gestational age (3.53, 1.14–5.92), birth length (–2.71, –4.90 to –0.53) and placental weight (2.58, 0.86–4.29) were independent predictors of serum adiponectin levels in subjects aged 4–5 years. Conclusion : Birth length, not birth weight, is an independent predictor of serum adiponectin levels in Japanese young children.