The Arya and Paris model (AP) predicts soil water retention curves from soil particle distribution (PSD) based on the distribution (PSD) between these two parameters using a scaling factor á to estimate porous radius from particle radius. In a previous work we have demonstrate that the best performance of the model is obtained using á as a function of the water content (è), obtained with a group of 104 samples of Brazilian soils. In the present work we have applied the Arya and Paris model for the UNSODA database, comparing the predicted soil water content with measured values of soil water retention curves for laboratory and field determinations. Both data set (field and laboratory) have shown good correlation (RMSD between 0.05 and 0.08 m3 m-3) between estimated by AP model and measured data, but a group of soil presented very bad correlation indicating possible imprecise retention or particle size distribution data in the database.