Probability distribution models (PDMs) of variables are cornerstones in the performance-based earthquake engineering framework. For liquefiable ground in earthquake-prone regions, the liquefiable layer often is sandwiched between nonliquefiable layers. However, PDMs for thicknesses of these soil layers, particularly the nonliquefiable crust and the liquefiable layer, are not yet well established. To fill this gap, this study leveraged field case history data from different regions to develop empirical PDMs for the soil-layer thicknesses of liquefiable ground. Seventeen probability density functions (PDFs) were examined to identify the appropriate ones that well characterize the thickness uncertainties. It was found that the lognormal PDF is an appropriate choice for either thickness variable at either region of interest. Means and standard deviations of the established PDMs for different regions, such as China, Japan, New Zealand, and the US, were tabulated for the expedience of implementation.