Eight competitive oarswomen (age, 22 +/- 3 years; mass, 64.4 +/- 3.8 kg) performed three simulated 2,000-m time trials on a rowing ergometer. The trials, which were preceded by a 24-hour dietary and training control and 72 hours of caffeine abstinence, were conducted 1 hour after ingesting caffeine (6 or 9 mg á kg-1 body mass) or placebo. Plasma free fatty acid concentrations before exercise were higher with caffeine than placebo (0.67 +/- 0.34 vs. 0.72 +/- 0.36 vs. 0.30 +/- 0.10 mM for 6 and 9 mg á kg-1 caffeine and placebo, respectively; p <.05). Performance time improved 0.7% (95% confidence interval [CI] 0 to 1.5%) with 6 mg á kg-1 caffeine and 1. 3% (95% CI 0.5 to 2.0%) with 9 mg á kg-1 caffeine. The first 500 m of the 2,000 m was faster with the higher caffeine dose compared with placebo or the lower dose (1.53 +/- 0.52 vs.1.55 +/- 0.62 and 1. 56 +/- 0.43 min; p =.02). We concluded that caffeine produces a worthwhile enhancement of performance in a controlled laboratory setting, primarily by improving the first 500 m of a 2,000-m row.