At exercise steady state, the lower the arterial oxygen saturation (SaO(2)), the lower the O(2) return (QvO(2)). A linear relationship between these variables was demonstrated. Our conjecture is that this relationship describes a condition of predominant sympathetic activation, from which it is hypothesized that selective beta1-adrenergic blockade (BB) would reduce O(2) delivery (QaO(2)) and QvO(2). To test this hypothesis, we studied the effects of BB on QaO(2) and QvO(2) in exercising humans in normoxia and hypoxia. O(2) consumption VO(2), cardiac output Q, CO(2) rebreathing), heart rate, SaO(2) and haemoglobin concentration were measured on six subjects (age 25.5 +/- 2.4 years, mass 78.1 +/- 9.0 kg) in normoxia and hypoxia (inspired O(2) fraction of 0.11) at rest and steady-state exercises of 50, 100, and 150 W without (C) and with BB with metoprolol. Arterial O(2) concentration (CaO(2)), QaO(2) and QvO(2) were then computed. Heart rate, higher in hypoxia than in normoxia, decreased with BB. At each VO(2), Q was higher in hypoxia than in normoxia. With BB, it decreased during intense exercise in normoxia, at rest, and during light exercise in hypoxia. SaO(2) and CaO(2) were unaffected by BB. The QaO(2) changes under BB were parallel to those in Q.QvO(2) was unaffected by exercise in normoxia. In hypoxia the slope of the relationship between QaO(2) and VO(2) was lower than 1, indicating a reduction of QvO(2) with increasing workload. QvO(2) was a linear function of SaO(2) both in C and in BB. The line for BB was flatter than and below that for C. The resting QvO(2) in normoxia, lower than the corresponding exercise values, lied on the BB line. These results agree with the tested hypothesis. The two observed relationships between QvO(2) and SaO(2) apply to conditions of predominant sympathetic or vagal activation, respectively. Moving from one line to the other implies resetting of the cardiovascular regulation.