Using RTV rubber as an interface between mirror elements and their supporting structures during grinding and polishing was proposed for the Advanced X-ray Astrophysics Facility (AXAF) for glass safety concerns. This paper shows that the mirror performance is quite sensitive to the compression modulus of GE RTV-60 which, like all other rubber-like materials, is very difficult to characterize by testing and even more difficult to characterize analytically. Consequently, using representative RTV properties in mirror analyses only produces nominal performance predictions. The envelope of the range of performance has to be determined by using both extremes of the RTV compression modulus. This paper also presents a comparison between compression moduli generated via testing and that from semi-empirical formulas. The agreement is satisfactory when mean values of the test data are used for comparison and the shape factor is modified to include partially constrained surfaces. The scatter ranges from 7.6% to 23.0% depending upon the way the RTV samples were cast. Using an error of about 16.2, we were still able to meet the error budget requirements for the Glass Support Fixture (GSF).