A practical problem for industries of different areas is the comparison of the durability of similar manufactured components produced by different sources under normal working conditions. Typically, the time and costs of such comparisons are long and expensive. Efficient industrial experiments for handle such comparisons consist of submitting components to accelerated tests. where stress levels of covariates are considered. In this paper, we develop a Bayesian analysis for comparing two Weibull regression models for the life times with a general stress-response model which includes some of the most used models in accelerated tests. The exponential life model is also discussed as a especial case of the Weibull model. A generated data set illustrate the methodology.