Abstract Background: Alveolar-capillary membrane diffusing capacity for carbon monoxide (DMCO) and pulmonary capillary volume (Vcap) can be estimated by the multi-step Roughton and Foster (RF, original method from 1957) or the single-step NO-CO double diffusion technique (developed in the 1980s). The latter method implies inherent assumptions. We sought to determine which combination of the alveolar membrane diffusing capacity for nitric oxide (DMNO) to DMCO ratio, an specific conductance of the blood for NO (θNO) and CO (θCO) gave the lowest week-to-week variability in patients with heart failure. Methods: 44 heart failure patients underwent DMCO and Vcap measurements on three occasions over a ten-week period using both RF and double dilution NO-CO techniques. Results: When using the double diffusing method and applying θNO = infinity, the smallest week-to-week coefficient of variation for DMCO was 10 %. Conversely, the RF method derived DMCO had a much greater week-to-week variability (2x higher coefficient of variation) than the DMCO derived via the NO-CO double dilution technique. The DMCO derived from the double diffusion technique most closely matched the DMCO from the RF method when θNO = infinity and DMCO = DLNO/2.42. The Vcap measured week-to-week was unreliable regardless of the method or constants used. Conclusions: In heart failure patients, the week-to-week DMCO variability was lowest when using the single-step NO-CO technique. DMCO obtained from double diffusion most closely matched the RF DMCO when DMCO/2.42 and θNO = infinity. Vcap estimation was unreliable with either method.