The objective of the study was to further unravel the prognostic significance of body weight changes in patients with COPD. Two survival analyses were performed: (1) a retrospective study, including 400 patients with COPD none of whom had received nutritional therapy; (2) a post hoc analysis of a prospective study, including 203 patients with COPD who had participated in a randomized placebo-controlled trial. There was no overlap between the patient groups. Baseline characteristics of all patients were collected on admission to a pulmonary rehabilitation center in stable clinical condition. In the prospective randomized placebo-controlled trial, the physiologic effects of nutritional therapy alone (n = 71) or in combination with anabolic steroid treatment (n = 67) after 8 wk was studied in patients with COPD prestratified into a depleted group and a nondepleted group. Mortality was assessed as overall mortality. The Cox proportional hazards model was used to quantify the relationship between the baseline variables age, sex, spirometry, arterial blood gases, body mass index (BMI), smoking, and subsequent overall mortality. Additionally, the influence of treatment response on mortality was investigated in the prospective study. The retrospective study revealed that low BMI (p < 0.001), age (p < 0.0001) and low PaO2 (p < 0.05) were significant independent predictors of increased mortality. After stratification of the group into BMI quintiles a threshold value of 25 kg/m2 was identified below which the mortality risk was clearly increased. In the prospective study, weight gain (> 2 kg/8 wk) in depleted and nondepleted patients with COPD, as well as increase in maximal inspiratory mouth pressure during the 8-wk treatment, were significant predictors of survival. On Cox regression analysis weight change entered as a time-dependent covariate remained an independent predictor of mortality in addition to all variables that were entered in the retrospective study. The combined results of the two survival analyses provide evidence to support the hypothesis that body weight has an independent effect on survival in COPD. Moreover the negative effect of low body weight can be reversed by appropriate therapy in some of the patients with COPD.