Obesity is a complex multifactorial disease involving genetic and environmental factors and influencing several different metabolic pathways. In this regard, metabonomics, that is the study of complex metabolite profiles in biological samples, may provide a systems approach to understand the global metabolic regulation of the organism in relation to this peculiar pathology. In this pilot study, we have applied a nuclear magnetic resonance (NMR)-based metabolomic approach on urinary samples of morbidly obese subjects. Urine samples of 15 morbidly obese insulin-resistant (body mass index>40; homeostasis assessment model of insulin resistance>3) male patients and 10 age-matched controls were collected, frozen and analyzed by high-resolution (1)H-NMR spectroscopy combined with partial least squares-discriminant analysis. Furthermore, two obese patients who underwent bariatric surgery (biliopancreatic diversion and gastric bypass, respectively) were monitored during the first 3 months after surgery and their urinary metabolic profiles were characterized. NMR-based metabolomic analysis allowed us to identify an obesity-associated metabolic phenotype (metabotype) that differs from that of lean controls. Gut flora-derived metabolites such as hippuric acid, trigonelline, 2-hydroxyisobutyrate and xanthine contributed most to the classification model and were responsible for the discrimination. These preliminary results confirmed that in humans the gut microflora metabolism is strongly linked to the obesity phenotype. Moreover, the typical obese metabotype is lost after weight loss induced by bariatric surgery.