Disability and exertional dyspnea associated with chronic obstructive pulmonary disease has led to the development of rehabilitation programmes that aim to increase exercise tolerance and relief of dyspnea. To evaluate whether aerobic training (training groups P1 and P4), strength training (P2 and P5) or a combination of both (P4 and P6) is useful, 69 patients (44 m/25 f) with moderate to severe COPD were randomised to a three week inpatient training program. The training consisted of three weekly twenty minute exercise sessions without (P1 - P3) or with supplemental oxygen (P4 - P6) on a calibrated ergocycle (70 % W(max)) or three weekly sessions of 20 - 25 repetitions of 2 - 4 training series (40 % W(max)) or a combination of both. In general, the programme failed to demonstrate significant changes in lung function and arterial blood gases. Evaluation of exercise capacity via the six-minute-walking test (6MT) yielded a significant increase of the walking distance in all groups except P2 (60 - 83 m), The time to finish a test-set of daily activities (TAF) was reduced in all groups (5 - 58 sec) and reached significance in P1, P3, P5 and P6. After the 6MT, exertional dyspnea improved in all groups except P4 and was significant in P1 and P3; after the TAF, dyspnea again was reduced in all groups with a significant change in P2 and P5. These data support the hypotheses that a short term inpatient training programme is suitable to improve exercise-capacity and dyspnea. Patients with advanced disease (P4 - P6) show greater benefits with strength training (alone or in combination with aerobic training) while for patients with moderate disease (P1 - P3) aerobic training is favourable. These changes may translate into improved performance of daily activities and general well-being.