Despite interest in competitive strategy by coaches and athletes, there are no systematically collected data regarding the effect of differences in pacing strategy on the outcome of middle distance (2-4 min duration) events. In this study different pacing strategies were evaluated using a 2-km time trial on a bicycle attached to a wind load simulator. Well-trained subjects (N = 9) performed five separate time trials with the pace during the first 50% of the trial experimentally constrained within the usual real world range from very slow (approximately 55% of best time) to very fast (approximately 48% of best time). Serial VO2 was measured to estimate the oxidative contributions to the trial and accumulated O2 deficit and postexercise blood lactate measured to estimate the anaerobic contribution to the trial. The evenly paced trial (first 1 km = 50.9% final time) produced the fastest total time. The starting pace to final time relationship was described by a U shaped second order polynomial curve with the nadir for final time at a starting pace of 51% of best total time. There were no systematic differences in serial VO2, accumulated O2 deficit, or postexercise lactate that could account for the pacing related variations in performance. The data support the concept of relatively even pacing in middle distance events with negative consequences for even small variations in this strategy.