The metabolic activity of the brain has not been evaluated during physical exercise. In six volunteers substrate uptake by the brain was determined during graded exercise and recovery from maximal exercise by measuring the arterial-internal jugular venous concentration differences(a-v differences). The a-v difference for lactate increased from 0.02 +/- 0.08 mmol l-1 at rest to 0.39 +/- 0. 13 mmol l-1 during exercise and remained positive during 30 min of recovery (P < 0.05). The a-v difference for glucose (0.55 +/- 0.06 mmol l-1 at rest) did not change significantly during exercise, but during the initial 5 min of recovery it increased to 0.83 +/- 0.10 mmol l-1 (P < 0.05). The O2 a-v difference at rest of 3.11 +/- 0.30 mmol l-1 remained stable during exercise, then increased during the initial 5 min of recovery (3.77 +/- 0.52 mmol l-1) and remained high during the subsequent 30 min recovery period (3.62 +/- 0.64 mmol l-1; P < 0.05). Thus the O2/glucose uptake ratio did not change during exercise (pre-exercise 5.95 +/- 0.68; post-exercise 6.02 +/- 1.39) but decreased to 4.93 +/- 0.99 during the initial 5 min of recovery (P < 0.05). When lactate uptake was included, the resting O2/carbohydrate uptake ratio of 5.84 +/- 0.73 was reduced to 4.42 +/- 0.25 during exercise and decreased further during the recovery phase (to 3.79 +/- 0.30; P < 0.05). In contrast, in the resting and immobilised rat, lactate infusion to a level similar to that obtained during maximal exercise in humans did not affect the a-v difference for lactate. The large carbohydrate uptake by the brain during recovery from maximal exercise suggests that brain glycogen metabolism is important in the transition from rest to exercise, since this would explain the significant post-exercise decrease in the O2/carbohydrate uptake ratio.