Pulmonary rehabilitation is recommended in international treatment guidelines for chronic obstructive pulmonary disease (COPD). No one has however studied the effect on long-term mortality. The aim of the current study was to study the mortality in a sample of patients with severe COPD included in a 1-year multidisciplinary rehabilitation program. Body composition was assessed at baseline using bioelectrical impedance. Mortality was studied in 86 patients using the Cox proportional hazards model. Forty-seven (55%) of the patients died during the mean follow-up time which was almost 6 years. Risk of mortality increased with increasing age, increasing number of hospital days the year before inclusion and men had higher mortality risk than women. The mortality risk decreased with increasing % reference body weight, increasing fat-free mass index (FFMI), increasing FEV₁ and increasing 6-min walking distance. Gender, age and FFMI continued to be statistical significant predictors of mortality when controlling for the other baseline variables in a multivariate analysis. To conclude, body composition, measured by bioelectrical impedance and presented as FFMI, is an independent predictor of mortality in COPD patients.