We tested the hypothesis that restricting either step frequency (SF) or stride length (SL) causes a decrease in ventilatory response with limited breath frequency during sinusoidal walking. In this study, 13 healthy male and female volunteers (mean ± SD; age: 21.5 ± 1.8 years, height: 168 ± 7 cm, weight: 61.5 ± 8.3 kg) participated. The walking speed was sinusoidally changed between 50 and 100 m⋅min–1 with periods from 10 to 1 min. Using a customized sound system, we fixed the SF at 120 steps⋅min–1 with SL variation (0.83–0.41 m) (SFfix) or fixed the SL at 0.7 m with SF variation (143–71 steps⋅min–1) (SLfix) during the subjects’ sinusoidal walking. Both the subjects’ preferred locomotion pattern without a sound system (Free) and the unprompted spontaneous locomotor pattern for each subject (Free) served as the control condition. We measured breath-by-breath ventilation [tidal volume (VT) and breathing frequency (Bf)] and gas exchange [CO2 output (CO2), O2 uptake (O2)]. The amplitude (Amp) and the phase shift (PS) of the fundamental component of the ventilatory and gas exchange variables were calculated. The results revealed that the SFfix condition decreased the Amp of the Bf response compared with SLfix and Free conditions. Notably, the Amp of the Bf response under SFfix was reduced by less than one breath at the periods of 5 and 10 min. In contrast, the SLfix condition resulted in larger Amps of Bf and E responses as well as Free. We thus speculate that the steeper slope of the E-CO2 relationship observed under the SLfix might be attributable to the central feed-forward command or upward information from afferent neural activity by sinusoidal locomotive cadence. The PSs of the E, O2, and CO2 responses were unaffected by any locomotion patterns. Such a sinusoidal wave manipulation of locomotion variables may offer new insights into the dynamics of exercise hyperpnea.