Abstract: The thermal trip of low-voltage DC circuit breakers are characterized by high reliability and long life, so there is little failure data available. In order to ensure the safe operation of low-voltage distribution systems, life prediction and reliability assessment of thermal breakers are required. Accelerated degradation test is carried out on the thermal release as an accelerating stress, and the pseudo-failure life of the thermal release is extrapolated by using the obtained small-sample degradation data, and its pseudo-failure life obeys the two-parameter Weibull distribution by K-S test. Then the life prediction model of the thermal disconnectors is established based on Bayes and its life prediction is carried out. The gray Markov model is used to expand the small-sample life data as a priori information; the parameters of the posterior distribution of the life prediction model are estimated by the Markov chain Monte Carlo (MCMC) algorithm based on Gibbs sampling. Finally, the lifetime of the thermal disconnectors under normal stress is introduced externally based on the Arrhenius accelerated model to be approximately 3 047 days.