The relative importance of ventilatory, circulatory and peripheral muscle factors in determining tolerance to exercise in patients with COPD is not known. In twelve COPD patients (FEV1 :58 ± 17%pred.) we measured ventilation, cardiac output, dynamic hyperinflation, local muscle oxygenation, blood lactate and time to exhaustion during a) interval exercise (IE) consisting of 30 s at 100% peak work-rate alternated with 30 s at 50% and b) constant load exercise (CLE) at 75% WRpeak, designed to produce the same average work rate. Exercise time was substantially longer during IE than CLE (19.5 ± 4.8 versus 11.4 ± 2.1 min, p = 0.0001). Total work output was therefore greater during IE than CLE (81.3 ± 27.7 versus 48.9 ± 23.8 kJ, p = 0.0001). Dynamic hyperinflation (assessed by changes from baseline in inspiratory capacity-ΔIC) was less during IE than CLE at CLE exhaustion time (isotime, p = 0.009), but was similar at exhaustion (ΔICCLE : -0.38 ± 0.10 versus ΔICIE : -0.33 ± 0.12 l, p = 0.102). In contrast, at isotime, minute ventilation, cardiac output and systemic oxygen delivery did not differ between protocols (p>0.05). At exhaustion in both protocols, vastus lateralis and intercostal muscle oxygen saturation were higher in IE than CLE (p = 0.014 and p = 0.0002, respectively) and blood lactate concentrations were lower (4.9 ± 2.4 mmol/l versus 6.4 ± 2.2 mmol/l, p = 0.039). These results suggest that 1) exercise tolerance in COPD is limited by dynamic hyperinflation; and 2) cyclically lower (50%) effort intervals in IE help preserve muscle oxygenation and reduce metabolic acidosis compared to CLE at the same average work rate, but these factors do not appear to determine time to exhaustion.