The relationship of glycogen availability to performance and blood metabolite accumulation during repeated bouts of maximal exercise was examined in 11 healthy males. Subjects performed four bouts of 30 s maximal, isokinetic cycling exercise at 100 rev.min-1, each bout being separated by 4 min of recovery. Four days later, all subjects cycled intermittently to exhaustion [mean (SEM) 106 (6) min] at 75% maximum oxygen uptake (VO2max). Subjects were then randomly assigned to an isoenergetic low-carbohydrate (CHO) diet [7.8 (0.6)% total energy intake, n = 6] or an isoenergetic high-CHO diet [81.5 (0.4)%, n = 5], for 3 days. On the following day, all subjects performed 30 min cycling at 75% VO2max and, after an interval of 2 h, repeated the four bouts of 30 s maximal exercise. No difference was seen when comparing total work production during each bout of exercise before and after a high-CHO diet. After a low-CHO diet, total work decreased from 449 (20) to 408 (31) J.kg-1 body mass in bout 1 (P < 0.05), from 372 (15) to 340 (18) J.kg-1 body mass in bout 2 (P < 0.05), and from 319 (12) to 306 (16) J.kg-1 body mass in bout 3 (P < 0.05), but was unchanged in bout 4. Blood lactate and plasma ammonia accumulation during maximal exercise was lower after a low-CHO diet (P < 0.001), but unchanged after a high-CHO diet. In conclusion, muscle glycogen depletion impaired performance during the initial three, but not a fourth bout of maximal, isokinetic cycling exercise. Irrespective of glycogen availability, prolonged submaximal exercise appeared to have no direct effect on subsequent maximal exercise performance.