Seven male subjects completed cycle exercise bouts to the limit of tolerance on three occasions: (1) at a constant work rate (340+/-57 W; even-pace strategy; ES); (2) at a work rate that was initially 10% lower than that in the ES trial but which then increased with time such that it was 10% above that in the ES trial after 120 s of exercise (slow-start strategy; SS); and, (3) at a work rate that was initially 10% higher than that in the ES trial but which then decreased with time such that it was 10% below that in the ES trial after 120 s of exercise (fast-start strategy; FS). The expected time to exhaustion predicted from the pre-established power-time relationship was 120 s in all three conditions. However, the time to exhaustion was significantly greater (P<0.05) for the FS (174+/-56 s) compared with the ES (128+/-21 s) and SS (128+/-30 s) conditions. In the FS condition, (.)VO2 increased more rapidly toward its peak such that the total O2 consumed in the first 120 s of exercise was greater (ES: 5.15+/-0.78; SS: 5.07+/-0.83; FS: 5.36+/-0.84 L; P<0.05 for FS vs ES and SS). These results suggest that a fast-start pacing strategy might enhance exercise tolerance by increasing the oxidative contribution to energy turnover and hence "sparing" some of the finite anaerobic capacity across the transition to high-intensity exercise.