Two-dimensional (2D) relaxation exchange nuclear magnetic resonance (NMR) is in many ways similar to 2D frequency exchange NMR, except that the encoding times are comparable to the exchange time. This fact prevents the straightforward analysis of the cross-peak intensities in terms of joint probability densities, and quantitative information and understanding can only be obtained by comparison with simulated spectra. Based on simulations, an explanation is proposed as to why interference between relaxation and exchange may lead to asymmetric 2D exchange maps when exchange occurs between more than two sites. Practically, retro-fitting a simulated data set to an experimental one is shown to allow for the determination of the experimental relaxation and exchange parameters. This point is illustrated by studying a two-site model system consisting of interstitial water exchanging within a pack of spherical silica particles.