The prevention of perinatal brain damage following preterm birth remains a public health priority. Melatonin has been shown to be a promising neuroprotectant in neonatal preclinical models of brain damage, but few studies have investigated melatonin secretion in newborns. We hypothesized that melatonin circulating levels would be lower in preterm compared to term infants. We conducted a prospective, longitudinal, multicenter study to assess melatonin, and 6-sulfatoxy-melatonin (aMT6s) concentrations, measured by radioimmunoassay. Among 209 neonates recruited, 110 were born before 34 gestational weeks (GW) and 99 born after 34 GW. Plasma melatonin concentrations, measured at birth and on Day 3 were below detectable levels (≤7 pg/mL) in 78% and 81%, respectively, of infants born before 34 GW compared to 57% and 34%, respectively, of infants born after 34 GW. The distribution of plasma melatonin concentrations was found to be correlated with gestational age at both time-points (p < 0.001). Median urine aMT6s concentrations were significantly lower in infants born before 34 GW, both on Day 1 (230 ng/L vs. 533 ng/L, p < 0.0001) and on Day 3 (197 ng/L vs. 359 ng/L, p < 0.0001). In conclusion, melatonin secretion appears very low in preterm infants, providing the rationale for testing supplemental melatonin as a neuroprotectant in clinical trials.