Central (intracerebral ventral) and peripheral (subcutaneous and intraperitoneal) administration of ghrelin causes obesity in rodents by increasing food intake and decreasing fat oxidation. Recent studies in humans have shown that plasma ghrelin concentration was inversely related to body fat and was lower in Pima Indians, a population susceptible to obesity. Whether ghrelin plays a role in the etiology of obesity in humans is unknown. We, therefore, measured plasma ghrelin concentration before and after two interventions in monozygotic twins previously studied at Laval University, Quebec City. Twelve pairs of monozygotic twins were overfed by 84,000 kcal over a 100-day period, whereas another seven pairs of monozygotic twins were submitted to a 53,000 kcal negative energy balance induced by exercise over a 93-day period. At baseline, for all the subjects, plasma ghrelin concentration was negatively correlated with body mass and body fatness (r varying from 0.36 to 0.45). The intraclass coefficient for the twin resemblance (r(I) = 0.75; p = 0.006) indicated that plasma ghrelin concentration is a familial trait. In response to the 100-day intervention, plasma ghrelin exhibited a non-significant decrease of 61 +/- 30 fmol/l (p = 0.18) with overfeeding and a non-significant increase of 58 +/- 34 fmol/l (p = 0.17) with negative energy balance. However, there was no relationship between baseline plasma ghrelin concentration and the magnitude of body weight change in both interventions. These first experimental data under "clamped energy balance conditions" do not provide evidence that plasma ghrelin is involved in the etiology of human obesity. However, studies in free-living individuals are needed to clarify this question.