Abstract The use of correlation matrices to evaluate the number of uncorrelated stirrer positions of electromagnetic reverberation chambers has widespread applications in electromagnetic compatibility. We present a review of recent methods based on multivariate correlation functions that relates statistical inhomogeneities in space (frequency) to the reduction of uncorrelated cavity configurations. Full wave finite-difference time domain simulations of an actual reverberation chamber are performed through an in-house parallel code. The efficiency of this code allows for capturing extensive inhomogeneous/anisotropic reverberation fields at frequencies close to the lowest usable frequency (LUF) of the chamber. The concept of effective independent position is revised in light of random sampling and a model-driven relation with the probability distribution of correlation matrix entries is used to take into account spatial (frequency) inhomogeneities. Driven by extensive simulation data, an empirical probability density function is found for the correlation matrix elements to be non-central t -student distributed with asymmetry increasing towards low frequencies.