The aim of this study was to assess the within subject reproducibility of cardioventilatory measurements and the maximum permitted 'normal' variability over time at rest and exercise using the respiratory mass spectrometer (RMS). Ten subjects underwent an incremental exercise test on three separate occasions utilising rebreathing (RB) and helium dilution mixed expired gas analysis (HME) functions of the RMS. Measurements included heart rate (HR), oxygen consumption (V(O2)), carbon dioxide excretion (V(VO2)), effective pulmonary blood flow (Q(eff)), stroke volume (SV), arteriovenous oxygen content difference (AVO), transfer factor (Dl(CO)), functional residual capacity (FRC), minute ventilation (VE), tidal volume (VT) and respiratory quotient (RQ). The coefficients of variation for each variable for the 10 subjects were calculated. At rest, the 90th centile variability for measured cardiopulmonary variables (RB only) was <35%. During exercise, the 90th centile for variability for measured cardiopulmonary variables for HME and RB were < or =20 and <40%, respectively. These measurements in healthy adults should inform sample size in research studies.