Background: Leptin is an adipocytokine that suppresses appetite, and levels of leptin are low in conditions of undernutrition such as anorexia nervosa (AN). Although leptin exhibits pulsatility, secretory characteristics of leptin have not been well described, particularly in adolescents and in AN. There is also evidence that leptin may play a role in the regulation of certain neuroendocrine pathways; however, the contribution of alterations in leptin secretion to the changes observed in levels of growth hormone (GH) and cortisol in AN has not been explored. Hypotheses: We hypothesized that low leptin levels in AN reflect a decrease in basal and pulsatile leptin secretion, and that leptin secretory characteristics may contribute to increased levels of GH and cortisol in AN. Methods: Leptin concentration and secretory characteristics were studied by analyzing data from frequent sampling every 30 minutes overnight for 12 hours (2000 to 0800 hours) in 23 girls with AN and 21 healthy controls 12-18 years old, using Cluster and deconvolution methods. We also examined GH, cortisol and ghrelin secretory dynamics, and fasting levels of other nutritionally regulated hormones such as follicle stimulating hormone (FSH), luteinizing hormone (LH), estradiol, thyroid stimulating hormone (TSH) and thyroid hormones (total T3 and T4). Body composition was assessed using dual energy x-ray absorptiometry. Results: On Cluster analysis, girls with AN had lower leptin area under the curve (AUC), mean and valley leptin than controls (p<0.0001 for all). Lower leptin concentrations were subsequent to decreased basal leptin secretion (p=0.02) and decreased secretory burst mass (p<0.0001), which resulted in decreased pulsatile and total leptin secretion (p<0.0001 for both). Approximate entropy did not differ between the groups. Leptin secretory characteristics were strongly predicted by markers of nutritional status including body mass index (BMI), % body fat and an insulin resistance index (HOMA-IR: homeostasis model of assessment). In a regression model that included BMI, % body fat and ghrelin AUC in addition to leptin parameters, leptin secretory and concentration characteristics independently predicted GH burst interval and burst frequency. Valley mean leptin contributed to 56.2% of the variability in GH burst interval. Basal leptin secretion and fasting ghrelin contributed to 26.4% and 16.2% of the variability of GH burst frequency. Leptin parameters were also independent predictors of urinary free cortisol normalized for creatinine (UFC/cr), and pulsatile leptin secretion contributed to 14.6% of the variability of UFC/cr. Valley mean leptin contributed to 21.5% of the variability of cortisol half-life. Pulsatile leptin independently predicted levels of estradiol (10.7% of variability), and leptin concentration predicted levels of total T3 and T4. Fasting leptin correlated more strongly with pulsatile than with basal leptin secretion, and the strongest correlation was observed between fasting leptin and mean leptin concentration (r=0.97, p<0.0001). Conclusion: Low leptin levels in AN are a consequence of decreased basal and pulsatile leptin secretion, and are nutritionally regulated. Leptin independently predicts GH burst frequency and burst interval, as well as cortisol half-life and UFC/cr. Ghrelin is also an independent predictor of GH burst frequency. Low leptin and high ghrelin levels may be dual stimuli for the high GH concentrations observed in undernutrition. In addition, leptin predicts levels of estradiol and thyroid hormones.