It is known that non-cachectic patients with chronic obstructive pulmonary disease (COPD) respond well to pulmonary rehabilitation, but whether cachectic COPD patients are capable of adaptive responses is both important and unknown. 10 cachectic and 19 non-cachectic COPD patients undertook high-intensity cycling training, at the same relative intensity, for 45 min x day(-1), 3 days x week(-1) for 10 weeks. Before and after rehabilitation vastus lateralis muscle biopsies were analysed morphologically and for the expression of muscle remodelling factors (insulin-like growth factor (IGF)-I, myogenic differentiation factor D (MyoD), tumour necrosis factor (TNF)-alpha, nuclear factor (NF)-kappaB and myostatin) and key components of ubiquitin-mediated proteolytic systems (muscle ring finger protein (MURF)-1 and Atrogin-1). Rehabilitation improved peak work-rate and the 6-min walk distance similarly in non-cachectic (18+/-3% and 42+/-13 m, respectively) and cachectic (16+/-2% and 53+/-16 m, respectively) patients, but quality of life only improved in non-cachectic COPD. Mean muscle fibre cross-sectional area increased in both groups, but significantly less in cachectic (7+/-2%) than in non-cachectic (11+/-2%) patients. Both groups equally decreased the proportion of type IIb fibres and increased muscle capillary/fibre ratio. IGF-I mRNA expression increased in both groups, but IGF-I protein levels increased more in non-cachectic COPD. MyoD was upregulated, whereas myostatin was downregulated at the mRNA and protein level only in non-cachectic patients. Whilst rehabilitation had no effect on TNF-alpha expression, it decreased the activation of the transcription factor NF-kappaB in both groups by the same amount. Atrogin-1 and MURF-1 expression were increased in cachectic COPD, but it was decreased in non-cachectic patients. Cachectic COPD patients partially retain the capacity for peripheral muscle remodelling in response to rehabilitation and are able to increase exercise capacity as much as those without cachexia, even if they exhibit both quantitative and qualitative differences in the type of muscle fibre remodelling in response to exercise training.