This paper presents an approach to develop degradation and reliability models of analog integrated circuit (IC) voltage regulators based on the long-term evolution of the electromagnetic compatibility (EMC) performance degradation due to the stress time-dependent accelerated degradation test (ADT). The ADT plan is designed and conducted on six samples of both UA78L05 and L78L05 ICs placed inside a climatic chamber combining both the thermal step-stress (i.e., 70-110 ∘C) and constant electrical overstress (i.e., 9 and 12 V) conditions for a total stress duration of 950 hours. All the selected UA78L05 and L78L05 samples are subjected to the direct power injection (DPI) measurement test under nominal conditions in order to characterize their immunity to electromagnetic interference (EMI). The statistical degradation data (i.e., the average injected power) of the aged samples is computed across the entire DPI frequency range for a variety of stress time duration. The proposed log-linear accelerated life-stress test (ALT) model is combined with the Weibull unreliability distribution function model to estimate the failure lifetime data against the applied voltage stress at three different failure threshold criterion. At various constant voltage overstress and threshold constraints, the lifetime reliability performance parameters (i.e., time-to-failure, probability of failure, model constants) of the tested device under tests (DUTs) were evaluated based on the measured degradation data. It is demonstrated that, for a limited number of samples under the combined influence of thermal step-stress with voltage overstress conditions, the proposed reliability model predicts with a very acceptable accuracy the lifetime reliability of both UA78L05 and L78L05 tested ICs, developed based on the conducted immunity degradation data. The physics-based modeling approach is utilized to develop the model for the degradation paths based on the observed monotonic degradation of the measured degradation data as well as the conditions of the thermal step-stress ADT. In order to estimate the unknown parameters of the developed degradation model, the maximum likelihood estimation (MLE) method is combined with a genetic optimisation algorithm.