This study examined the effects of pre-exercise carbohydrate availability on the time to exhaustion for moderate and heavy exercise. Seven men participated in a randomized order in two diet and exercise regimens each lasting 3 days with a 1-week interval for washout. The tests were performed at 50% of the difference between the first (LT1) and second (LT2) lactate breakpoint for moderate exercise (below LT2) and at 25% of the difference between the maximal load and LT2 for heavy exercise (above LT2) until exhaustion. Forty-eight hours before each experimental session, subjects performed a 90-min cycling exercise followed by 5-min rest periods and a subsequent 1-min cycling bout at 125% VO2max/1-min rest periods until exhaustion to deplete muscle glycogen. A diet providing 10% (CHO(low)) or 65% (CHO(mod)) energy as carbohydrates was consumed for 2 days until the day of the experimental test. In the exercise below LT2, time to exhaustion did not differ between the CHO(mod) and the CHO(low) diets (57.22 +/- 24.24 vs 57.16 +/- 25.24 min). In the exercise above LT2, time to exhaustion decreased significantly from 23.16 +/- 8.76 min on the CHO(mod) diet to 18.30 +/- 5.86 min on the CHO(low) diet (P < 0.05). The rate of carbohydrate oxidation, respiratory exchange ratio and blood lactate concentration were reduced for CHO(low) only during exercise above LT2. These results suggest that muscle glycogen depletion followed by a period of a low carbohydrate diet impairs high-intensity exercise performance.