AimsIn this multicentre study, we compared cardio-pulmonary exercise test (CPET) parameters between heart failure (HF) patients classified as New York Heart Association (NYHA) class I and II to assess NYHA performance and prognostic role in mild HF. Methods and resultsWe included consecutive HF patients in NYHA class I or II who underwent CPET in three Brazilian centres. We analysed the overlap between kernel density estimations for the per cent-predicted peak oxygen consumption (VO 2 ), minute ventilation/carbon dioxide production (VE/VCO 2 ) slope, and oxygen uptake efficiency slope (OUES) by NYHA class. Area under the receiver-operating characteristic curve (AUC) was used to assess the capacity of per cent-predicted peak VO 2 to discriminate between NYHA class I and II. For prognostication, time to all-cause death was used to produce Kaplan-Meier estimates. Of 688 patients included in this study, 42% were classified as NYHA I and 58% as NYHA II, 55% were men, and mean age was 56 years. Median global per cent-predicted peak VO 2 was 66.8% (IQR 56-80), VE/VCO 2 slope was 36.9 (31.6-43.3), and mean OUES was 1.51 (±0.59). Kernel density overlap between NYHA class I and II was 86% for per cent-predicted peak VO 2 , 89% for VE/VCO 2 slope, and 84% for OUES. Receiving-operating curve analysis showed a significant, albeit limited performance of per cent-predicted peak VO 2 alone to discriminate between NYHA class I vs. II (AUC 0.55, 95% CI 0.51-0.59, P = 0.005). Model accuracy for probability of being classified as NYHA class I (vs. NYHA class II) across the spectrum of the per cent-predicted peak VO 2 was limited, with an absolute probability increment of 13% when per cent-predicted peak VO 2 increased from 50% to 100%. Overall mortality in NYHA class I and II was not significantly different (P = 0.41), whereas NYHA class III patients displayed a distinctively higher death rate (P < 0.001). Conclusions Patients with chronic HF classified as NYHA I overlapped substantially with those classified as NYHA II in objective physiological measures and prognosis. NYHA classification may represent a poor discriminator of cardiopulmonary capacity in patients with mild HF.