Inverse Laplace transform techniques are commonly used to convert nuclear magnetic resonance (NMR) relaxation data in porous media into lifetime distributions. In this connection, we present a direct comparison of T2 measurements made on monosized packings of sintered glass beads and numerical simulations based on the grain consolidation model. Three systems, with porosities of 0.38, 0.22, and 0.14, were studied, and in each case we found that the position and width of the measured T2 distribution was well represented by the random walk simulations. These systems are generally in the fast diffusion (weak surface relaxation) regime and have relatively narrow T2 distributions. For this reason, we have found that inverse Laplace transform techniques are particularly sensitive to random noise in the measured and simulated decay spectra.