PURPOSE: To determine whether a dose-response relationship exists between caffeine and 2000-m rowing performance. METHODS: In this randomized, placebo-controlled, double-blind crossover study, 10 competitive male rowers (mean +/- SD: age = 20.6 +/- 1.4 yr, body mass = 87.7 +/- 10.5 kg, height = 186.8 +/- 6.8 cm, (.)VO2peak = 5.1 +/- 0.6 L x min(-1)) consumed 2, 4, or 6 mg x kg(-1) caffeine or a placebo 60 min before completing a 2000-m time trial on a rowing ergometer. The trials were preceded by a 24-h standardized diet (including a light preexercise meal of 2 g x kg(-1) CHO), and subjects were tested preexercise for hydration, caffeine abstinence, and blood glucose concentrations. RESULTS: Time trial performance was not significantly different across the three caffeine doses or placebo (P = 0.249). After the three caffeine trials, postexercise plasma glucose and lactate concentrations were higher compared with the placebo trial (P < 0.05). Plasma caffeine concentrations after 60 min of ingestion were lower than the values reported previously by others following the same dose, and there was considerable interindividual variation in plasma caffeine concentrations in response to the various caffeine doses.