Previous work has shown that replacing N2 in air with He at the same inspired O2 fraction reduces the exercise-induced alveolar-arterial PO2 difference (AaPO2) in horses but has provided no mechanism explaining this effect. We sought to distinguish among possible causes by using the multiple inert gas elimination technique. Six horses were studied on a high-speed treadmill while they breathed either ambient air or normoxic He-O2. O2 uptake reached 138.0 ml.min-1.kg-1 and was not affected by He-O2. Temperature-corrected arterial PO2 was 76.7 Torr (air) and 86.9 Torr (He-O2) (P < 0.01). Corresponding AaPO2 was 22.3 and 15.9 Torr, respectively (P < 0.01). Mean AaPO2 predicted from ventilation-perfusion inequality did not change with He-O2 (12.7 Torr with air and 11.9 Torr with He-O2). Mean arterial PCO2 was 50.1 Torr with air and 44.1 Torr with He-O2 (P < 0.01); minute ventilation and tidal volume were correspondingly higher by 140 l/min and 1.0 liter, respectively, with He-O2. Pulmonary O2 diffusing capacity, cardiac output, and all ventilation-perfusion dispersion indexes did not change with He-O2. Intrapulmonary shunt was insignificant. Higher ventilation with He-O2 explained only approximately 4 Torr of the 10-Torr rise observed in arterial PO2. The remainder (and the corresponding fall in AaPO2) was due to more complete diffusion equilibration as a consequence of the higher minute ventilation and thus alveolar PO2, which reduced the average slope of the O2 dissociation curve, thereby increasing the ratio of diffusive to perfusive conductance.