OBJECTIVE: To determine whether fetal leptin levels correlate with fetal weight and whether such correlation is direct or indirect via insulin or human placental lactogen (hPL), respectively. DESIGN: Cross-sectional study of offspring at term (n=175) with over-representation of large-for-gestational age (LGA; n=70) and small-for-gestational age (SGA; n=23) cases in a population of Caucasian women with no pregnancy pathology. METHODS: Fetal cord blood was collected after delivery. In several cases (n=62) paired mother-fetus blood samples were obtained. Leptin, insulin and hPL levels were measured by RIA. Anthropometric data (birth weight, body mass index, placental weight) were recorded. RESULTS AND CONCLUSIONS: Maternal insulin, hPL and leptin levels were higher than fetal concentrations. Cord blood leptin levels positively correlated with the anthropometric data with stronger correlations in female (0.54<r<0.66) than in male (0.32<r<0.39) neonates. Cord blood leptin levels did not differ between appropriate-for-gestational age (AGA; n=82) and SGA (n=23) neonates, but were higher (P<0.001) by 83% in LGA (n=70) than in AGA neonates. Among the different weight classes the correlations between fetal leptin and anthropometric data were only observed in LGAs, but not in AGAs or SGAs. Fetal, but not maternal, leptin levels strongly correlated with fetal insulin (r=0.56; P<0.001). After accounting for this close relationship insulin could no longer be used to predict birth weight (r=0.15, P=0.051). We suggest that the correlation of cord blood insulin with neonatal weight in LGAs is, in addition to insulin's direct anabolic action, indirectly mediated via leptin. It is hypothesized that fetal insulin stimulates fetal adipocyte leptin production.