Endurance training increases muscle content of glucose transporter proteins (GLUT-4) but decreases glucose utilization during exercise at a given absolute submaximal intensity. We hypothesized that glucose uptake might be higher in trained vs. untrained muscle during heavy exercise in the glycogen-depleted state. Eight untrained subjects endurance trained one thigh for 3 wk using a knee-extensor ergometer. The subjects then performed two-legged glycogen-depleting exercise and consumed a carbohydrate-free meal thereafter to keep muscle glycogen concentration low. The next morning, subjects performed dynamic knee extensions with both thighs simultaneously at 60, 80, and until exhaustion at 100% of each thigh's peak workload. Glucose uptake was similar in both thighs during exercise at 60% of thigh peak workload. At the end of 80 and at 100% of peak workload, glucose uptake was on average 33 and 22% higher, respectively, in trained compared with untrained muscle ( P < 0.05). Training increased the muscle content of GLUT-4 by 66% ( P < 0.05). At exhaustion, glucose extraction correlated significantly ( r = 0.61) with total muscle GLUT-4 protein. Thus, when working at a high load with low glycogen concentrations, muscle glucose uptake is significantly higher in trained than in untrained muscle. This may be due to the higher GLUT-4 protein concentration in trained muscle.