The swelling pressure of compacted bentonite under complex thermo-chemical conditions is a critical safety parameter of the engineering barrier system for a deep geological repository. To investigate the cyclic thermal and saline effects on the swelling pressure of densely compacted Gaomiaozi (GMZ) bentonite, a series of constant-volume swelling pressure tests were performed using deionized water and NaCl solutions. At 20 °C and 60 °C, the multi-step salinization-desalinization process generated a hysteresis of the swelling pressure, and multiple saline cycles led to its diminishing attenuation. In the framework of the double structure theory, the observed hysteresis of the swelling pressure was ascribed to the irreversible collapse of inter-aggregate pores under osmotic pressure, while the gradual attenuation was attributed to the further homogenization of the clay structure. Soaking tests were also conducted at various thermal and saline conditions to provide evidence for the cyclic saline effects. No salt precipitation was identified due to the negligible changes in the total suctions of the specimens after saline cycles conducted at 20 °C and 60 °C. The membrane effect (resistance of compacted clay to salt migration) was considered, and its degree reflected the changes in the osmotic suctions of the specimens occurred after the soaking tests. The resistance of compacted GMZ bentonite to salt migration was relatively strong during wetting with low saline solutions (0–0.5 mol/L) at 20 °C, and became weak during wetting with high saline solutions (1.0–2.0 mol/L) or at 60 °C. The variation of the swelling pressure during thermal cycling depended on the maximum temperature and concentration of the presaturation solution. By assuming the validity of the membrane effect at the maximum temperatures (valid at 40 °C and invalid at 60 °C), the inconsistent changes in the swelling pressure during thermal cycling were explained by considering the incomplete initial saturation and role of thermal osmosis. However, further verification of the proposed hypothesis by X-ray diffraction and membrane testing at elevated temperatures is required.