We compared the effect of bilateral knee joint cooling with or without a pre-cooling warm-up on sprint cycling performance to a non-cooling control condition. Seventeen healthy young males (25 ± 2 years, 174 ± 6 cm, 70 ± 9 kg) performed three conditions in a counterbalanced order (condition 1: warming + cooling + cycling; condition 2: cooling + cycling; condition 3: cycling). For warming, a single set of cycling intervals (a 10 s sprint with maximal effort followed by a 180 s active recovery; resistive load 4% and 1% body mass for sprint and recovery, respectively) was performed. For cycling, five sets of cycling intervals were performed. For cooling, 20 min of bilateral focal knee joint cooling was applied. Peak and average values of anaerobic capacity and wheel cadence during each set across conditions were statistically compared. There was no condition effect over set (condition × set) in anaerobic capacity (F8,224 < 1.49, p > 0.16) and wheel cadence (F8,224 < 1.48, p > 0.17). Regardless of set (condition effect: F2,224 > 8.64, p < 0.0002), conditions 1 and 2 produced higher values of anaerobic capacity (p ≤ 0.05). Similarly (condition effect: F2,224 > 4.62, p < 0.02), condition 1 showed higher wheel cadence (p < 0.02) than condition 3. A bilateral joint cooling for 20 min with or without pre-cooling warm-up may improve overall sprint cycling capacity during five sets of cycling intervals when compared to the non-cooling condition.