In humans, EIAH severity correlates most consistently and inversely witha-a Do 2. Interindividual differences in PaCO2PaCO2 or the ventilatory equivalent forV˙o 2 (orV˙co 2) are also commonly found to correlate significantly with EIAH; however, there are many exceptions, especially in mild EIAH, and thus the degree of hyperventilation accounts for less of the variance in PaO2PaO2 in most studies. Those men and women, in both young and old age groups, who experience severe EIAH have almost equal contributions from the absence of hyperventilation and wideneda-a Do 2to their hypoxemia, compared with nonhypoxemic subjects at comparableV˙o 2 max. For combined human and animal group mean data (Table 1), variations in O2 saturation at maximum exercise are best predicted from a multiple linear-regression model (r = 0.93)2, where ventilation (as reflected by PaCO2PaCO2 ) explains ∼60% of the variance in SaO2SaO2 ,V˙o 2 max accounts for 25% of it, anda-a Do 2for the remainder.