Metabolic exercise testing is valuable in patients with chronic heart failure (CHF), but periodic breathing may confound the measurements. We aimed to examine the effects of periodic breathing on the measurement of oxygen uptake ( V*O(2)) and respiratory exchange ratio (RER). First, we measured the effects of different averaging procedures on peak V*O(2) and RER values in 122 patients with CHF undergoing cardiopulmonary exercise testing. Secondly, we studied the effects of periodic breathing on V*O2) and RER in healthy volunteers performing computer-guided periodic breathing. Thirdly, we used a Fourier analysis to study the effects of periodic breathing on gas exchange measurements. The first part of the study showed that 1 min moving window gave a mean peak V*O(2) of 13.8 ml.min(-1).kg(-1) for the CHF patients. A 15 s window gave significantly higher values. The difference averaged 1.0 ml.min(-1).kg(-1) ( P <0.0001), but varied widely: 41% of subjects showed a difference greater than 1.0 ml.min(-1).kg(-1). RER values were also higher by an average of 0.09 ( P <0.0001); in 20% of subjects the difference was greater than 0.10. In the second part of the study, we found artefactual elevations of peak V*O(2) (without averaging) of 2.9 ml.min(-1).kg(-1) ( P <0.01) and of peak RER of 0.13 ( P <0.001), which were still significant when 30 s averaging was applied [delta(peak V*O(2))=1.8 ml.min(-1).kg(-1), P <0.01; deltaRER=0.08, P <0.001]. The third, theoretical, part of the study showed that values of carbon dioxide output and V*O(2) oscillate with different phases and amplitudes, resulting in oscillations in their ratio, RER. Averaging over 15 s or 30 s can be expected to give only 10% or 36% attenuation respectively. Thus periodic breathing causes variable artefactual elevations of measured peak V*O(2) and RER, which can be attenuated by using longer averaging periods. Clinical reports and research publications describing peak V*O(2) in CHF should be accompanied by details of the averaging technique used.