Background: Muscle strength imbalance assessment (e.g., hamstring-to-quadriceps ratio, H:Q) using traditional peak torque isokinetic measurements has been shown to be a weak risk factor predictor of future lower-limb injuries (e.g., hamstring strain and anterior cruciate ligament tear). In soccer, power-related tasks are commonplace and injuries are most likely to occur during fatigued high-velocity actions. Thus, it is reasonable to that calculating H:Q using power output and may serve as an alternative to traditional peak torque-based H:Q. Aims: We aimed to investigate the relationship of isokinetic H:Q calculated from traditional peak torque and power output during non-fatigue and fatigue conditions. Methods: Seventy-nine professional soccer players (25.6 ± 4.9 years old; 78.7 ± 8.1 kg; 179.4 ± 6.7 cm) performed concentric knee extension-flexion contractions at 60°.s−1 (five repetitions) and 300°.s−1 (30 repetitions, fatigue trial). Traditional peak torque H:Q was calculated using the highest torque obtained during five repetitions at 60°.s−1. Power output H:Qnon-fatigued was calculated using the average from the 2nd, 3rd and 4th repetitions, and power output H:Qfatigued was obtained as the average of the power output of the last three repetitions of the fatigue trial. Results: Weak (rs = 0.27) and moderate (rs = 0.49) correlations were found between traditional peak torque and power output H:Qfatigued, and traditional peak torque and power output H:Qnon-fatigued, respectively. Conclusion: The present data suggested that power H:Q differs from traditional H:Q, particularly during fatigue in professional soccer players, which warrants further investigation on the potential use of power output H:Q ratios for injury prediction.