Purpose Repeated sprint training in hypoxia (RSH) performed with voluntary hypoventilation at low lung volumes (RSH-VHL) has been proposed as an alternative to RSH. Recently, repeated sprint training combined with end-expiratory breath hold until the breaking point (RSH-UBP) has been suggested as a more effective stimulus. The aim of this study was to compare the physiological responses of repeated sprint training in normoxia (RSN), RSH-VHL and RSH-UBP. We hypothesized that RSH-UBP would induce greater hypercapnic and hypoxemic stress compared to the other conditions and therefore greater physiological responses. Methods Ten healthy active men participated in three repeated sprint exercise sessions on ergocycle. Pulse oxygen saturation (SpO2), end-tidal CO2 pressure (PETCO2), muscle oxygenation, cardiac hemodynamics, and total work were continuously monitored throughout the protocol. Results Time spent below 96% SpO2 was higher in both RSH-VHL and RSH-UBP than in RSN (82 ± 66 s and 74 ± 57 s vs. 11 ± 14 s, P = 0.039). Contradictory, PETCO2 was higher in RSH-VHL and RSH-UBP than in RSN (33.6 ± 4.2 and 32.9 ± 3.5 vs. 27.8 ± 4.2 mmHg, p textless 0.001). Total work was lower in RSH-UBP than in RSH-VHL and RSN (33.1 ± 5.8 vs. 46.5 ± 5.7 and 47.4 ± 6.0 kJ, p textless 0.001) despite a higher mean power in RSH-UBP. Amplitude in the muscle deoxygenation-reoxygenation cycle was lower during RSH-UBP (p textless 0.045). There was no difference in stroke volume or cardiac output between the three conditions. Conclusion Overall, RSH-VHL emerged as the most effective condition for inducing large hypercapnic and hypoxemic stimuli while maintaining the training load during repeated sprint training, whereas RSH-UBP succeeded in inducing similar hypercapnic and hypoxic stress with shorter sprint durations.