To determine the mechanism by which low arterial PO2 (PaO2) affects muscle carbohydrate (CHO) metabolism during exercise, dogs inhaled gas consisting of 0.21 (NO; n = 6) or 0.11 (LO; n = 6) inspired oxygen fraction (FIO2) during rest and 150 min of moderate treadmill exercise. Limb arteriovenous difference and isotopic ([3H]- and [14C]glucose) methods were used to assess muscle carbohydrate metabolism: PaO2 was reduced by approximately 50% in LO vs. NO, but limb O2 uptake was similar. Glucose disappearance was increased during rest (13 +/- 2 vs. 19 +/- 1 mumol.kg-1.min-1) and exercise (23 +/- 4 vs. 36 +/- 6 mumol.kg-1.min-1 at 150 min) in LO vs. NO, but arterial glucose was unchanged because hepatic glucose production was increased similarly. Limb glucose and pyruvate oxidation (derived from vein [14C]lactate specific activity) rates were elevated about twofold during rest and exercise in LO vs. NO. Estimated limb glycogenolysis increased at rest (21 +/- 9 vs. 96 +/- 23 mumol/min) and during exercise (70 +/- 21 vs. 184 +/- 41 mumol/min at 150 min) in LO vs. NO. The %CO2 and %lactate from glucose in LO were about twofold the values in NO in rest and exercise. The %CO2 from pyruvate was greater and free fatty acid levels were lower, suggesting reduced fat metabolism in LO. Arterial lactate and pyruvate levels were elevated during rest and the initial 30 min of exercise, even though net limb outputs were no greater. Lactate-to-pyruvate ratios and pH were similar in LO and NO during exercise.(ABSTRACT TRUNCATED AT 250 WORDS)