Our objective was to study the post-training response to exercise, comparing fatigue-limited (FL) vs. dyspnea-limited (DL) COPD patients. Moderate and severe COPD patients (GOLD definition) were included. They were classified as FL if Borg score of fatigue at maximal exercise testing was ≥ 2 points vs. dyspnea; and DL if it was the reverse. Also, each patient was evaluated with submaximal cycloergometry, 6 minutes walking test and quality of life score (SGRQ). All patients were trained 3 times/week, 90 min/session with aerobic and strength exercises by 8 weeks. A total of 14 patients in LF and 11 in LD group were evaluated with same tools. Means of age were 69 and 66 years respectively. They presented severe airway flow obstruction (FEV1: 49%). There was not any baseline difference between both groups, except body-mass index, which was lower in FL. Both groups significantly improved p exercise variables post-training in comparison with baseline and SGRQ, except maximal workload in DL. Comparing both groups, FL had the highest maximal workload (48.7 ± 9.2 vs. 40.04 ± 15.48 watts, p = 0.033), 6 minute walking test (505.42 ± 50.75 vs. 454.9 ± 64.3 meters, p = 0.048) and endurance time (14.57 ± 9.55 vs. 6.71 ± 4.18 min, p = 0.025), respectively. It can be concluded that FL patients had better response after training in maximal and submaximal exercise tests in comparison with DL. Perhaps, different training strategies would be performed to train different COPD phenotypes.