Door armrests of different crush properties and placement were evaluated in a series of side impact sled tests. Three armrest designs were fabricated with an identical shape but different crush force. The crush properties covered a range in occupant protection systems based on knowledge of human tolerance in side impacts. With BioSID, the softest armrest produced the lowest compression and Viscous responses, and the probability of AIS 4+ injury was below 1%. The compression-based responses increased significantly in tests with armrests of a higher crush force. The profile of the stiffer armrests clearly protruded into the dummy, and the probability of serious injury was 86%-100% based on compression. With SID, the lowest TTI(d) was with the intermediate stiffness armrest. The SID dummy and TTI(d) criterion indicated a 4%-8% probability of AIS 4+ injury for all test conditions and armrest designs. Spearman's rank correlation coefficient was rs = 0.09 indicating a difference in sorting designs with VC (or C) in BioSID and TTI(d) in SID. In companion tests, AIS 4 injuries occurred in anesthetized animals exposed to similar side impact loadings from the intermediate stiffness armrest, whereas AIS 2 injuries occurred with the softest design. These data confirm the appropriateness of VC (or C) in BioSID and insensitivity of TTI(d) in SID for side impact injury assessment.