To determine if decreased systemic oxygen (O2) extraction contributes to the exercise limit in severe chronic obstructive pulmonary disease (COPD), 40 consecutive incremental cycle ergometer exercise tests performed by such patients, from which a "log-log" lactate threshold (LT) was identified, were compared to those of 8 patients with left ventricular failure (LVF) and 10 normal controls. Pulmonary gas exchange and minute ventilation were measured continuously and arterial blood gas tensions, pH, and lactate concentrations were sampled each minute. Cardiac output (Qc) was measured by first-pass radionuclide ventriculography. The systemic O2 extraction ratio (O2ER) was calculated as arterial - mixed venous O2 content difference (CaO2 - CvO2)/CaO2. Peak exercise O2 uptake (VO2peak) was markedly reduced in both COPD and LVF [41 (3) and 42 (3)% predicted, respectively], compared to controls [89 (2)% predicted, P < 0.0001 for each]. Similarly, the LT occurred at a low percentage of predicted maximal oxygen consumption in both COPD and LVF [25 (2) and 27 (3)%] compared to normals [46 (3)%, P < 0.0001 for each]. The systemic O2ER at peak exercise was severely reduced in COPD [0.36 (0.02)] compared to the other groups [P < 0.0001 for each], for whom it was nearly identical [0.58 (0.03) vs 0.63 (0.04), LVF vs control, P > 0.05]. In the COPD group, an early LT correlated with reduced systemic O2ER at peak exercise (r = 0.64, P < 0.0001), but not with any index of systemic O2 delivery. These data suggest that lactic acidemia during exercise in patients with severe COPD is better related to abnormal systemic O2 extraction than to its delivery and contributes to the exercise limit.