In previous research, physiological resilience has been measured as deterioration of the physiological profile during prolonged exercise. This study aimed to evaluate the test–retest reliability of physiological resilience during prolonged moderate-intensity running. Physiological profile of 26 well-trained endurance runners (10 females) was tested in nonfatigued state as well as during and after two identical ∼2.5-h long physiological resilience tests at ∼89% of VT1 (ventilatory threshold 1) speed within an average period of 13 days. Reliability was assessed with intraclass correlation coefficients (ICC), coefficient of variations (CV%), and typical errors (TE and TE%). Change in maximal speed achieved in the incremental test (sPeak) was the most reliable metric and showed good reliability (ICC: 0.81 and TE: 1.8). The drifts in running economy (RE) and the heart rate (HR) had moderate to good reliability in the second half of the trial (ICC: 0.52–0.80 and TE 1.7–2.4). The changes in maximal oxygen uptake (VO_2max) and ventilatory thresholds (VTs) had poor reliability (ICC: 0.07–0.36 and TE: 3.5–4.6). However, the absolute values of these variables demonstrated good to excellent reliability in fatigued state (ICC > 0.83, TE% < 5.2%, and CV% < 4.1%) even though they significantly deteriorated. Determining changes in sPeak and drifts in RE and HR appears to be the most reliable method to measure physiological resilience. In contrast, measurement of the physiological profile remains reliable in fatigued state for every variable. Longer or more demanding protocols may be required to obtain greater reliability for deterioration of the physiological profile.