The effect of different dosages of caffeine (0 - 5 - 9 - 13 mg · kg body weight-1) on endurance performance was examined. Nine well-trained cyclists participated in this study (VO2max 65.1 + 2.6 ml · kg-1 · min-1). Caffeine capsules were administered in random order and double-blind. One hour after capsule ingestion, subjects cycled until exhaustion at 80 % Wmax on an electromagnetically braked cycle ergometer. Blood samples were taken before, during and after the exercise test. Before and after the test a urine sample was obtained. A significant increase in endurance performance was found for all caffeine tests compared to placebo (endurance time 47 + 13, 58 ± 11, 59 ± 12 and 58 ± 12 min for 0, 5, 9 and 13 mg kg-1 body weight, respectively). No differences were found in endurance performance between the three caffeine dosages which indicates that no dose-response relation of caffeine and endurance performance was found. An increased free fatty acid and glycerol concentration was found after caffeine consumption compared with placebo. The mean urinary caffeine concentrations after exercise were 4.8 + 1.8, 8.9 ± 5.2 and 14.9 ± 6.9 μg ml-1 urine for 5, 9 and 13 mg of caffeine kg-1 body weight. Only the lowest dose of caffeine resulted in urine caffeine concentrations below the doping limit of the International Olympic Committee of 12 μg ml-1 urine in all individuals. It is concluded that caffeine is an ergogenic aid that stimulates endurance performance. A dose-response relation between caffeine and endurance time was not found for the dose-range investigated. The stimulating effect of caffeine was already apparent at the lowest dose of caffeine given (5 mg kg-1). At this dose urinary caffeine concentration remained below the doping limit in all subjects.