Accelerated life tests are used to obtain timely information on the durability and reliability of materials. Test units are subjected to higher than usual levels of “stress” and a model is used to estimate life at use conditions. Although it is desirable to use a physically-based model to justify the required extrapolation, in many practical situations, no such model is available or the physical basis for extrapolation is uncertain. In such situations, extrapolation is based on an empirical model. Sensitivity analysis tools then become important to assess the effect of model error and to allow engineers to make safe design decisions. This paper presents models, methods, and a description of software tools for performing systematic sensitivity analysis to assess potential model error. These methods are illustrated with an experiment that was conducted to determine if the fatigue life of a spring would meet a given specification.