Changes in expired alveolar O2 and CO2 were measured breath-by-breath in six healthy male subjects (mean age 30 yr, mean weight 80 kg) at rest, 600 kpm/min, and 1,200 kpm/min. Changes were expressed in relation to expired volume (liters) and time (s) and separated into an initial dead-space component using the Fowler method applied to expired CO2 and O2, and alveolar slope. The alveolar slopes with respect to time (dPACO2, dPAO2, Torr/s) increased in relation to CO2 output (VCO2, 1/min, STPD) and O2 intake (VO2, 1/min, STPD) but were reduced by increasing tidal volume (VT, liters, BTPS): dPACO2 = 2.7 + 4.6(VCO2) - 1.9(VT) (r = 0.97); and dPAO2 = 2.3 + 5.5(VO2) - 1.9(VT) (r = 0.96). From the alveolar slopes, tidal volume, and airway dead-space volume, mean expired alveolar PO2 and PCO2 (PAO2, PACO2) were calculated. There was no change in arterialized capillary PCO2 (PaCO2) between rest (38.9 +/- 0.66 Torr) and heavy exercise (38.2 +/- 2.18 Torr), but mean PACO2 rose from 36.7 +/- 0.55 to 40.8 +/- 1.67 Torr during heavy exercise. There was no change in arterialized capillary (mean = 84.3 +/- 0.7 Torr) or alveolar (mean = 107.2 +/- 1.03 Torr) PO2. Exercise increases the fluctuations in alveolar gas composition leading to discrepancies between the PCO2 in mean alveolar gas and arterial blood to an extent that is dependent on VCO2 and VT.