Periodic breathing (PB) during exercise in patients with congestive heart failure (CHF) is associated with prominent osclliations (OSC) of O2 uptake (VO2). We hypothesized that the VO2 OSC represent OSC in true O2 exchange, resulting from concomitant cardlac output fluctuations and are not merely due to OSC of lung O2 stores. We compared the amplitude of the OSC of VO2, ventilation (VE), and end-expiratory lung volume (EELV) in 17 patients with CHF and PB and in seven healthy control subjects who volitionally simulated PB. Subjects underwent an incremental and/or a constant work-rate exercise test. VE and VO2 were measured breath by breath. EELV change was estimated by summing the difference between inspiratory and expiratory tidal volumes for each breath. The amplitude of the OSC, DELTA, is expressed as the ratio of the difference between the peak and nadir of the oscillating variable divided by its mean [DELTA = (peak - nadir)/mean]. In CHF, during incremental testing, the amplitude of the VE OSC was smaller than that of the VO2 OSC (DELTA-VE = 49 +/- 15% [SD], DELTA-VO2 = 63 +/- 25%, p < 0.01). In contrast, during volitional PB in the control subjects, VE OSC were larger than VO2 OSC (DELTA-VE = 48 +/- 12%, DELTA-VO2 = 25 +/- 11%, p < 0.01). This suggests that changing VE itself cannot account for the marked VO2 OSC seen in CHF. In the patients, EELV showed no systematic OSC, did not correlate with DELTA-VO2, and was not significantly different from zero. In contrast, in the control subjects, EELV tended to be lower at peak VE (-0.14 +/- 0.5 L, p = 0.06), and it was well correlated with DELTA-VO2 (r = 0.6, p < 0.01). In conclusion, VO2 oscillations in patients with CHF are more prominent than the associated VE OSC, are greater than the VO2 OSC that can be induced by volitional PB, and are not associated with changes in EELV. This supports the hypothesis that cardlac output OSC contribute to VO2 OSC.