During the operation of artificial underground structures, the surrounding rock experiences fatigue and creep damage caused by several types of disturbances under long-term constant loading. To quantify the mechanical response of sandstone under creep–fatigue loading, a damage–hardening evolution model based on the linear superposition concept is proposed. In the model, coupling is applied to represent the synergistic effect of creep and fatigue. Creep–fatigue tests of sandstone specimens are conducted under multilevel loading. The damage and hardening effects of sandstone under creep–fatigue loading are complex. Hardening is the dominant effect under low creep–fatigue loads, and damage is the dominant effect under high creep–fatigue loads. The strength of the rock specimens undergoes increasing and decreasing trends under this loading path, and the evolution of the Mohr–Coulomb envelope is discussed. The proposed model can be used to describe the test data and the evolution of the creep–fatigue process. With increasing creep–fatigue number, the acoustic emission amplitude, energy, and cumulative counts increase. However, the amplitude is more sensitive than the energy, indicating that it is more suitable for describing creep–fatigue loading. Furthermore, the peak frequencies of the AE signals are mostly distributed in the 0–15 kHz, 15–30 kHz, 30–45 kHz, and 45–55 kHz regions. The signal proportion in the 45–55 kHz zone decreases with the creep–fatigue number. However, other frequency zones increase with the creep–fatigue number. This phenomenon illustrates that the crack scale of the specimens increases with the creep–fatigue number.