Cross-frames must provide necessary stiffness for stability, but excessive stiffness can be detrimental by causing stress distributions differing from designers’ expectations and / or uneconomical cross-frame designs. Cross-frames are also generally assumed to have important roles in live-load distribution. This work evaluates a novel means for considering the overall stress distribution throughout finite element models, in contrast to other contemporary methods that quantitatively analyze only isolated stresses. This method is used to evaluate different cross-frame designs, including the absence of cross-frames, for simply-supported composite skewed steel I-girder bridges. Considering all results, from both newly proposed and conventional metrics, it is demonstrated that the influence of cross-frame design often has a more significant influence than the presence or absence of cross-frames. Thus, the importance of cross-frames may be less than previously thought. Furthermore, the data reveals that the greatest differences between models were generally due to the use of staggered cross-frames, which also cause the most significant departures from traditional expectations of girder behavior.