Steel twin box-girders are considered an attractive option for the construction of bridges due to their basic design, simple form, and ease of creation. Despite their advantages, they are considered fracture critical and as such there is an additional mandate for these bridges to inspected more in depth. This causes their inspection cost to be approximately two to five times greater than that of bridges with non-fracture critical members. The required additional inspection in the U.S. is mainly driven by rare historical events of bridge collapse for bridges that were not steel twin box girders. In addition, the mandated additional inspection does not reflect the inherent level of redundancy in most bridges. Therefore, it is important to quantify the potential for fracture and the level of redundancy in steel two-girder bridges in general, and twin box girders in particular, to minimize their inspection cost. Recognizing the inherently large scatter in fatigue performance, evaluating crack propagation and potential for fracture should, however, be performed in a probabilistic manner using detailed models that represent accurate behavior of the bridge. In this study, a detailed numerical finite element model of steel twin tub-girder bridge is developed and crack growth analysis, potential for fracture of its main tubs, and its overall redundancy is evaluated. The crack growth analysis is performed using multi-mode linear elastic fracture mechanics while accounting for uncertainties in the random variables associated with crack propagation and fracture. The results of the crack growth analysis are utilized to develop fragility functions that specify inspection intervals versus probability of failure where failure is characterized by dynamic crack growth. The analysis conducted to quantify the potential for fracture show distinct possible failure modes that vary from brittle fracture to ductile fracture. The extreme loading case shows that the bridge overall is not at risk of collapse. It is important to note that this conclusion cannot be generalized for all tub girder bridges since the level of redundancy is expected to vary between bridges depending on many factors such as girders geometries, plate thickness, fabrication, among others. However, the presented approach and the corresponding results provide a systematic way by which fracture critical bridges can be evaluated.