Exercise critical power (CP) has been shown to represent the highest sustainable work rate (WR) in patients with chronic obstructive pulmonary disease (COPD). Parameter estimation, however, depends on 4 high-intensity tests performed, on different days, to the limit of tolerance (T(lim)). In order to establish a milder protocol that would be more suitable for disabled patients, we contrasted CP derived from 4, 3 and 2 tests (CP4, CP3 and CP2) in 8 males with moderate COPD. In addition, CP was calculated from 2 single-day tests performed on an inverse sequence (CP(2AB) and CP(2BA)): CP values within 5 W from CP4 were assumed as "clinically-acceptable" estimates. We found that [CP4-CP3] and [CP4-CP2] differences were within 5 W in 8 and 6 patients, respectively (95% confidence interval of the differences = -1.3 to 3.5 W and -11.5 to 6.5 W). There was a systematic decline on T(lim) when an exercise bout was performed after a previous test on the same day (P<0.05). Consequently, substantial differences were found between CP4 and any of the CP estimates obtained from single-day tests. In conclusion, clinically-acceptable estimates of CP can be obtained by using 3 or, in most circumstances, 2 constant WR tests in patients with moderate COPD--provided that they are not performed on the same day.