We compared the intraobserver and interobserver agreement of blood (BGT) and gas exchange (GET) methods for determination of the anaerobic threshold (AT) in patients with COPD. In addition, we determined the sensitivity and specificity of the gas exchange methods for determination of the AT. Two noninvasive methods, the V-slope (VS) and the ventilatory equivalents method (VEM) were compared with two blood sampling methods, the log standard HCO3 (SB) vs log VO2 (SBT) and base excess (BE) vs VO2 (BET). Twenty-nine patients with COPD (FEV1 < 60%) performed incremental exercise tests to exhaustion while breath-by-breath gas exchange measurements were made. Blood samples were drawn at the end of each minute for SB and BE. Two trained observers determined the VO2 at the threshold for each of the four indices on two separate occasions two weeks apart. Our results demonstrated the following: only modest interobserver and intraobserver agreement was noted by Spearman rank correlations; the VEM was as sensitive as the VS in COPD patients; and the presence of a true metabolic acidosis was not reliably predicted by GET methods. Moreover, although the blood methods accurately identified the presence of metabolic acidosis, there was disagreement on the actual point of the BGT. We conclude that gas exchange indices were not helpful for the determination of metabolic acidosis in patients with COPD.