Nutritional support can increase body weight and physiologic function in COPD, but there are some patients who do not respond to nutritional therapy. The aim of this prospective study was to describe the nonresponse to 8 wk of oral nutritional supplementation therapy (500 to 750 kcal/d extra), implemented in an inpatient pulmonary rehabilitation program, with respect to lung function, body composition, energy balance, and systemic inflammatory profile in 24 (16 male) depleted patients with COPD. On the basis of the weight change after 8 wk, patients were divided into three groups (Group 1: weight gain < 2% of baseline body weight, n = 5; Group 2: weight gain 2 to 5%, n = 9; Group 3: weight gain >/= 5%, n = 10). Although no differences were seen in lung function and body composition, Group 1 was characterized by older age, a lower baseline dietary intake/resting energy expenditure (REE) ratio, and a greater number of users of continuous supplemental oxygen when compared with Group 3. In addition, Group 1 exhibited higher baseline concentrations of fasting glucose and LPS-binding protein than did Groups 2 and 3. The concentrations of the soluble TNF- receptors 55 and 75 were elevated in Groups 1 and 2 when compared with Group 3. Furthermore, a significant, inverse correlation coefficient between baseline dietary intake and soluble intercellular adhesion molecule was revealed (r = -0.50, p = 0.016). On linear regression analysis, age, baseline intake/REE ratio, sTNF-receptor 55, and extracellular/intracellular water (ECW/ICW) ratio were selected as independent, significant parameters contributing to a total explained variation of 78% in weight change after nutritional therapy. In conclusion, nonresponse to nutritional therapy in COPD is associated with ageing, relative anorexia, and an elevated systemic inflammatory response. Further research is needed to investigate whether these factors contribute to eventual disturbances in intermediary metabolism as reflected by the increased glucose concentration and ECW/ICW ratio.