Abstract: Under conditions of varying amplitude low-cycle fatigue loading, materials have shown a pronounced load sequence effect during the fatigue damage accumulation process. This observed phenomenon introduces complexities in predicting the fatigue life under such varying amplitude conditions, making it a pressing challenge in the field of material science. Taking into account the concept of intrinsic damage dissipation—which establishes a one-to-one correspondence with fatigue damage and encapsulates the thermodynamic essence of low-cycle fatigue failures—we have meticulously derived an evolution model for intrinsic damage dissipation, termed as the type D model. This derivation is grounded in the principles of continuum damage mechanics coupled with the framework of irreversible thermodynamics. By adopting the equivalent intrinsic damage dissipation as a pivotal criterion for damage conversion, we have innovatively constructed a fatigue life prediction model for varying amplitude low-cycle loads, which inherently considers the load sequence effect. To ascertain the robustness, effectiveness, and advanced capabilities of our proposed model, we embarked on rigorous uniaxial low-cycle fatigue testing. This involved materials such as P355NL1 structural steel and the Ti-6Al-4V titanium alloy, subjected to two distinct stages of varying amplitude loads. Our empirical findings were then juxtaposed with established models in the domain, namely the Manson model, Kwofie model, and Peng model. The outcomes of our research unequivocally indicate that the predictions rendered by our novel model consistently fall within a margin of 1.5 times the error range. This level of accuracy is not only in close alignment with our experimental results but also demonstrably superior to the predictions offered by existing models. The application of the intrinsic damage dissipation theory, as showcased in our research, heralds a pioneering direction for predicting the fatigue life of metallic materials under varying amplitude conditions.