The scheduled replacement of the 8th North Bridge, in Salt Lake City, UT, presented a unique opportunity to test a pre-cast concrete deck, steel girder bridge. A live-load test was performed under the directions of Bridge Diagnostic Inc (BDI) and Utah State University. Six different load paths were chosen to be tested. The recorded data was used to calibrate a finite-element model of this superstructure, which was created using solid, shell, and frame elements. A comparison between the measured and finite-element response was performed and it was determined that the finite-element model replicated the measured results within 3.5% of the actual values. This model was later used to obtain theoretical live-load distribution factors, which were compared with the AASHTO LRFD Specifications estimations. The analysis was performed for the actual condition of the bridge and the original case of the bridge, which included sidewalks on both sides. The comparison showed that the code over predicted the behavior of the actual structure by 10%. For the original case, the code's estimation differed by as much as 45% of the theoretical values. Another opportunity was presented to test the behavior of a cast-in-place concrete box girder bridge in Joaquin County, CA. The Walnut Grove Bridge was tested by BDI at the request of Utah State University. The test was performed with six different load paths and the recorded data was used to calibrate a finite-element model of the structure. The bridge was modeled using shell elements and the supports were modeled using solid elements. The model was shown to replicate the actual behavior of the bridge to within 3% of the measured values. The calibrated model was then used to calculate the theoretical live-load distribution factors, which allowed a comparison of the results with the AASHTOO LRFD Specifications equations. This analysis was performed for the real conditions of the bridge and a second case where intermediate diaphragms were not included. It was determined that the code's equations estimated the behavior of the interior girder more accurately for the second model (within 10%) than the real model of the bridge (within 20%).