Abstract: Particle-size distribution (PSD) is the most fundamental soil physical property,which has a distinct impact on other physical properties such as soil hydraulic properties.As an input parameter for various simulation models,the accurate description of PSD is of vital importance to the reliability of simulation results.In the present study,five empirical models,i.e.,the log-normal distribution,the cubic spline,the logistic growth,the modified logistic growth and the van Genuchten models,are employed to describe the cumulative particle-size distribution data from laboratory experiments,an international database and literature publications.Results indicate that optimal agreements between measured and predicted values are obtained through the modified logistic growth model while the most frequently used log-normal model produces relatively poor agreements.Furthermore,the other three empirical models are in between the aforementioned two models in terms of prediction accuracy.Similar results are obtained in additional application of these models to estimate soil water retention characteristics using a fractal approach.