The assessment of chemoreflex sensitivity in heart failure patients is gaining increasing interest since recent studies demonstrated that augmented chemosensitivity is an independent predictor of mortality and represents an important pathogenic factor in the development of Cheyne-Stokes respiration. The single-breath CO2 test is a well-established method to quantify peripheral hypercapnic chemoreflex sensitivity. As the original criteria for the computation of the chemoreflex sensitivity in healthy subjects need to be modified in heart failure patients to take into account impaired cardiac function, the effects of such modifications on measurement reliability deserve investigation. Hence, we devised this study to assess the reliability of the single-breath CO2 test in heart failure patients. In 27 clinically stable, mild-to-moderate heart failure patients (age (mean±SD): 64±10 years, left ventricular ejection fraction: 34±7%, NYHA class: 2.7±0.4), the test was administered on two consecutive days in the same conditions. Reliability was assessed by the standard error of measurement (SEM) and by the intraclass correlation coefficient (ICC). The mean value of the chemoreflex sensitivity on the two days was: 0.25 ± 0.12 and 0.24 ± 0.12 l min(-1) mmHg(-1) (p = 0.45), respectively. The SEM was 0.05 l min(-1) mmHg(-1), indicating large intra-subject variability. Consequently, in order to be 95% confident that a real change has occurred between two measurements taken on the same individual (test-retest), the observed difference must be higher than ±0.15 l min(-1) mmHg(-1), which is about 60% of the mean value across our population. The ICC was 0.71, indicating thatintra-subject variability, although high, is a limited (29%) portion of inter-subject variability. Intra-subject variability should be carefully taken into account when using the single-breath CO2 test in assessing changes in individual patients. The observed ICC indicates that this test may provide useful information for diagnostic/classification purposes.