Ghrelin is an orexigenic peptide that is also a growth hormone (GH) secretagogue. Secretory dynamics of ghrelin have not been characterized in adolescent girls with anorexia nervosa (AN). We hypothesized that girls with AN would have increased ghrelin concentrations measured over a 12 hour period of nocturnal sampling compared with healthy adolescents as a consequence of increased basal and pulsatile ghrelin secretion, and that endogenous ghrelin secretory and concentration parameters would independently predict GH and cortisol. We performed overnight frequent sampling q 30 minutes over 12 hours in 22 girls with AN and 18 healthy adolescents 12-18 years old, and examined ghrelin concentration and secretory dynamics using Cluster and deconvolution analyses. Associations between ghrelin and GH, IGF-I, cortisol, leptin, total T3, luteinizing hormone (LH) and measures of insulin resistance were examined. Girls with AN were followed for one year and examined again at weight recovery (10% increase in BMI). On Cluster analysis, adolescent girls with AN had significantly higher ghrelin concentrations than healthy adolescents, including total area under the curve (AUC) for ghrelin (p=0.002), nadir ghrelin (p=0.0006) and valley ghrelin (p=0.002). On deconvolution analysis, secretory burst amplitude (p=0.03) and burst mass (p=0.04) were higher in girls with AN than in controls, resulting in higher pulsatile (p=0.05) and total ghrelin secretion (p=0.03) in AN. Basal ghrelin secretion and secretory burst frequency did not differ. Total ghrelin AUC predicted basal GH secretion (r=0.49, p=0.002), as did ghrelin burst mass, pulsatile and total ghrelin secretion. On multiple regression analysis, ghrelin concentration and secretion parameters were independent predictors of basal GH secretion. Fasting ghrelin concentration was an independent predictor of GH burst frequency (r=0.44, p=0.005). The nutritional markers BMI and percent body fat predicted post glucose ghrelin levels (r= -0.41, p=0.01; and r= -0.43, p=0.01) and valley ghrelin (r= -0.46, p=0.02, r= -0.43, p=0.03) but not fasting ghrelin. Total ghrelin AUC was inversely associated with fasting insulin (r= -0.50, p=0.002) and HOMA-IR (r= -0.54, p=0.0007). Ghrelin parameters were also inversely associated with leptin and IGF-I. Insulin resistance was the most significant predictor of all ghrelin parameters. Ghrelin concentration was strongly and positively associated with cortisol concentration, and valley ghrelin was an independent predictor of cortisol burst frequency accounting for 52% of the variability. Ghrelin was also an independent predictor of total T3 and LH levels. We conclude that higher ghrelin concentrations in adolescent girls with AN are a consequence of increased secretory burst mass and amplitude resulting in increased pulsatile and total ghrelin secretion. The most important predictor of ghrelin concentration is the degree of insulin resistance, and ghrelin in turn predicts basal GH secretion, GH secretory burst frequency and cortisol burst frequency.