We examined the relation between nocturnal melatonin and hormonal markers of puberty in 57 normal children and adolescents and 39 subjects with disorders of pubertal onset. Melatonin was measured in hourly blood samples drawn overnight by constant withdrawal. Basal 08.00 h plasma testosterone, estradiol and LH, and the peak LH response to LHRH administration were determined. There were no significant correlations between testosterone, estradiol, basal LH and peak LH and melatonin peak (r = -0.18, -0.22, -0.02, -0.12, respectively) or melatonin peak time (r = 0.12, -0.01, -0.02, 0.07 respectively). The results were not affected significantly by sex, diagnosis or age. A comparison of subjects grouped by peak LH < 15 U/l (most likely prepubertal; n = 40) and peak LH > 30 U/l (most likely pubertal; n = 34) showed no significant differences in melatonin peak (160.5 +/- 59.3 vs. 146.6 +/- 50.9 pg/ml; t = 1.09; p > 0.05) or melatonin peak time (1.8 +/- 1.7 vs. 2.5 +/- 1.7 h; t = -1.79; p > 0.05). Although a pineal-puberty relation cannot be excluded, the results do not support the hypothesis that melatonin restrains the hypothalamic-pituitary-gonadal axis during childhood.