Abstract: Continental bedded salt formations in China are characterized by abundant interlayers and pronounced heterogeneity, and cyclic gas injection and withdrawal can induce deformation incompatibility and local instability at soft-hard rock interfaces. To clarify the effects of layer thickness ratio on fatigue damage and interface failure in interlayered salt-rock composites, pure rock-salt specimens and composite specimens containing thin and thick gypsum interlayers were prepared using high-purity rock salt and gypsum from a representative mining area, with the bedded salt formation in the Dawenkou Basin, Tai'an, serving as the engineering background. The layer thickness ratio, λ, was defined as the ratio of gypsum interlayer thickness to total specimen height, and three groups with λ= 0, 0.2, and 0.4 were tested. Triaxial cyclic loading-unloading tests were conducted under a confining pressure of 10 MPa at a frequency of 0.4 Hz. Stress-strain responses, energy partitioning, a cumulative-dissipated-energy-based damage model, X-ray three-dimensional micro-computed tomography, and pore network modeling were integrated to characterize interface instability in terms of macroscopic deformation, energy evolution, and pore-fracture connectivity. The results show that increasing λ markedly suppresses the axial rheological deformation and lateral dilation of the rock-salt matrix. At a normalized cycle count of 0.5, the cumulative axial strain of the λ= 0.4 specimen was 1.74%, approximately 30% lower than the 2.44% measured for the λ= 0 specimen, while the steady-state cyclic creep rate decreased from 1.14 × 10⁻⁴ to 4.31 × 10⁻⁵ per cycle. In the late fatigue stage, its circumferential strain increased abruptly to −5.84%, while the volumetric strain reversed to -5.00%, indicating a transition from coordinated progressive deformation to sudden dilation following constrained compaction. The single-cycle energy response of all specimens shifted from dissipation-dominated to elastic-energy-storage-dominated behavior. As λ increased, the corresponding axial-strain threshold advanced successively from 0.75% to 0.55% and 0.45%, while the energy dissipation ratio remained below 0.05 during the stable stage. A damage variable based on cumulative dissipated energy and a layer-thickness-ratio-dependent damage acceleration factor were introduced. The coefficients of determination, R², exceeded 0.998 for all fitted curves, and the damage acceleration factor β increased from 3.71 to 4.87, demonstrating that interface constraint intensifies local energy storage and accelerates damage evolution. After fatigue loading, the volume fractions of pores and fractures larger than 20 nm in the specimens with λ= 0, 0.2, and 0.4 were 2.94%, 67.3%, and 79.3%, respectively. In the thin-interlayer specimen, enrichment bands around the upper and lower interfaces accounted for 13.00% of the CT slices, with a strong-enrichment coefficient of 5.82. As λ increased from 0.2 to 0.4, the mean coordination number of the pore network increased from 4.90 to 5.81, the mean pore radius increased from 14.83 to 26.75 nm, and the mean throat length increased from 68.30 to 159.32 nm. These changes indicate that dispersed small-scale pore-fracture connections progressively evolve into larger, well-connected pathways. Increasing the layer thickness ratio strengthens stiffness mismatch and interface constraint, causing fatigue instability to evolve through a sequence of constrained deformation, energy accumulation, and interface coalescence. Therefore, thick-interlayer interfaces should be regarded as critical zones for operational control of pressure fluctuations, cavern-wall monitoring, and long-term stability assessment of salt-cavern gas storage facilities.