There is considerable debate about the existence of a real eccentric overload in flywheel exercises. This study aimed to analyse the differences in concentric: eccentric mechanical output ratios between different loads and variables in the flywheel squat exercise. Twenty physically active men (22.9±2.2 years, height: 1.8±0.1 m, weight: 79.6±8.2 kg) performed a loading test using five moments of inertia. Angular speed was measured using a rotary encoder, while the vertical force was measured using force plates. For each variable (angular speed, angular acceleration, power, vertical force, and torque), mean and peak values were calculated for concentric and eccentric phases to allow comparisons across the loads. We tested the possible differences in Load×Phase (concentric and eccentric) and Load×Variable. The level of significance was established as p < 0.05. A significant Load×Phase interaction was found in mean angular speed, peak vertical force, peak angular acceleration, peak power and peak torque. Higher eccentric overload values were observed with speed-derived variables (angular speed, angular acceleration and power). In conclusion, speed-derived peak variables and lower loads are more likely to show an eccentric overload and can be used to monitor responses to flywheel training.