Introduction Central nervous system excitability depends on the task performed, muscle group solicited, and contraction type. However, little is known on corticospinal and motoneuronal excitability measured during locomotor exercise. This study aimed at determining the reliability of motor-evoked potentials (MEP) and thoracic motor-evoked potentials (TMEP) in dynamic mode during unfatiguing and fatiguing cycling exercise. Methods Twenty-two participants completed four visits. Visit 1 comprised familiarization and an incremental cycling test to determine maximal power output ( W max ). The remaining visits encompassed unfatiguing evaluations, which included a total of eight brief bouts of moderate- (50% W max ) and high-intensity cycling (80% W max ), four at each intensity. In each bout, a set of two TMEPs, five MEPs, and one M-max were obtained. Subsequently, a fatiguing exercise to exhaustion at 80% W max was performed, with four sets of measurements 3 min through the exercise and four additional sets at exhaustion, both measured at 50% W max . Results Intraclass correlation coefficients (ICCs) for 5, 10, 15, and 20 MEP·Mmax −1 revealed excellent reliability at both intensities and during cycling to exhaustion (ICC ≥0.92). TMEP·Mmax −1 showed ICCs ≥0.82 for moderate and high intensity, and it was not affected by fatigability. Overall standard error of measurement was 0.090 (0.083, 0.097) for MEP·Mmax −1 and 0.114 (0.105, 0.125) for TMEP·Mmax −1 . A systematic bias associated with the number of stimulations, especially at high intensity, suggested that the evaluation itself may be influenced by fatigability. A mean reduction of 8% was detected in TMEP·Mmax −1 at exhaustion. Conclusions Motoneuronal and corticospinal excitability measured in dynamic mode presented good to excellent reliability in unfatiguing and fatiguing exercise. Further studies inducing greater fatigability must be conducted to assess the sensitivity of central nervous system excitability during cycling.