To determine whether a single carbohydrate feeding could rapidly restore and maintain plasma glucose availability late in exercise, six trained cyclists were studied on two occasions during exercise to fatigue at 70 +/- 1% of VO2max. After 135 min of exercise, the men were fed either an artificially sweetened placebo or glucose polymers (3 g.kg-1 in a 50% solution). Prolonged exercise led to a decline in plasma glucose from 4.6 +/- 0.1 mM at rest to 3.9 +/- 0.2 mM after 135 min (P less than 0.05). Plasma glucose decreased further (P less than 0.05) to 3.2 +/- 2.0 mM at fatigue following placebo ingestion but increased (P less than 0.05) and was then maintained at 4.5-4.9 mM following carbohydrate ingestion. Respiratory exchange ratio (R) fell gradually during the placebo trial from 0.88 +/- 0.01 after 10 min of exercise to 0.81 +/- 0.01 at fatigue (P less than 0.01). R also reached a minimum of 0.81-0.82 in each subject after 135-180 min of exercise during the carbohydrate feeding trial but increased again to 0.84-0.86 as plasma glucose rose following the carbohydrate feeding. Exercise time to fatigue was 21% longer (205 +/- 17 vs 169 +/- 12 min; P less than 0.01) during the carbohydrate ingestion trial. Plasma insulin did not increase significantly, whereas plasma free fatty acids and blood glycerol plateaued following carbohydrate ingestion. These data indicate that a single carbohydrate feeding late in exercise can supply sufficient carbohydrate to restore euglycemia and increase carbohydrate oxidation, thereby delaying fatigue.