The presence of small amounts of residual void space (on the order of one percent by volume) is sufficient to make estimates of composite elastic constants (obtained while ignoring this porosity) violate the Hashin-Shtrikman bounds in many cases. Experimental values failing below the incorrectly estimated lower bound is the normal direction of violation, but erroneous neglect of porosity in ＇＇pure＇＇ samples can also lead to experimental values rising above the incorrectly computed upper bound. Examples are presented to illustrate the nature of the error and methods of correcting it for data on composite ceramics. The result is improved agreement between the Hashin-Shtrikman bounds, effective medium theory estimates, and experimental data.