Selected surface reverberation and bottom reverberation returns were used to compute estimates of the probability density function of the instantaneous reverberation. To estimate the densities, 6500 samples of surface reverberation and 3078 samples of bottom reverberation were used. The collections of samples were tested for randomness, independence, homogeneity, and normality. Both the surface and bottom reverberation were found to be non‐Gaussian. Kernel density estimation techniques were applied to the collections of samples to provide univariate estimates of the densities. The densities were seen to be nearly Gaussian, but with heavier tails. Heavier tailed densities generally result in higher false alarm rates for detectors designed for a Gaussian noise process.