While most performance metrics of high‐explosive (HE) based devices like detonation velocity, detonation pressure, and energy output are expected to degrade over time, the evolution of initiation threshold appears less clear, with claims of both increasing and decreasing trends in threshold having been made in the literature. This work analyzes D‐optimally designed sequential binary test data for a few thermally conditioned porous‐powder and polymer‐bonded HE initiator systems using a Bayesian likelihood method employing the probit regression model. We find that in most cases the initiation threshold decreases (i. e., sensitivity increases) upon accelerated thermal conditioning. However, such results are nuanced and influenced by factors like the contact area of initiating stimulus, HE characteristics like density and specific surface area, as well as possible thermally induced changes to other materials and interfaces involved.