In this paper, several probabilistic models for evaluating the fatigue performance of post-weld treated tubular bridge structures are briefly described. The application of the models is then demonstrated on a full-scale example bridge. Under realistic traffic loading conditions, it is shown that a significant increase in fatigue performance can be obtained with post-weld treatment. In comparing several treatment strategies, a partial treatment strategy is presented, which results in the same fatigue performance improvement as full treatment with a fraction of the treatment effort. One potential concern with the use of post-weld treatment methods such as needle peening is that they may simply shift the critical crack site to an untreatable location such as the weld root. To address this concern, a deterministic verification of this fatigue detail is presented. Shortcomings of this verification are then discussed, and future research needs highlighted.