This study was undertaken to determine whether carbohydrate feeding during exercise can delay the development of fatigue. Ten trained cyclists performed two bicycle ergometer exercise tests 1 wk apart. The initial work rate required 74 +/- 2% of maximum O2 consumption (VO2 max) (range 70-79% of VO2 max). The point of fatigue was defined as the time at which the exercise intensity the subjects could maintain decreased below their initial work rate by 10% of VO2 max. During one exercise test the subjects were fed a glucose polymer solution beginning 20 min after the onset of exercise; during the other they were given a placebo. Blood glucose concentration was 20-40% higher during the exercise after carbohydrate ingestion than during the exercise without carbohydrate feeding. The exercise-induced decrease in plasma insulin was prevented by carbohydrate feeding. The respiratory exchange ratio was unchanged by the glucose feeding. Fatigue was postponed by carbohydrate feeding in 7 of the 10 subjects. This effect appeared to be mediated by prevention of hypoglycemia in only two subjects. The exercise time to fatigue for the 10 subjects averaged 134 +/- 6 min (mean +/- SE) without and 157 +/- 5 min with carbohydrate feeding (P less than 0.01).