The most accurate critical velocity (CV) estimate for the prediction of velocity during a simulated 2,000-m rowing race and the relationship to aerobic power were studied. Sixteen male rowers completed randomized maximal exertion trials (200, 400, 600, 800, 1,000, and 1,200 m), a maximal oxygen consumption (VO(2)max) on a Concept II rowing machine, and an actual 2,000-m simulated rowing race. Three mathematical models were applied to 4 rowing distance combinations producing 12 CV estimates. Seven of the 12 possible CV estimates were not significantly different from actual 2,000-m velocity. Comparison of the 3 CV models using all 6 trial distances revealed that the nonlinear model produced a CV estimate lower than the 2 linear CV models. CV was significantly correlated to VO(2)max (r = 0.91) and the mean velocity achieved during the 2,000-m simulated rowing race (r = 0.97). VO(2)max was significantly correlated to 2,000-m simulated rowing race velocity (r = 0.93).