Equations were developed to predict soil matric potential as a function of soil water content, texture and bulk density in sandy soils. The equations were based on the additivity hypothesis - that water-retention of a whole soil depends on the proportions of several particle size fractions, each with fixed water-retention characteristics. The new model is an advancement over previously published models in that it embodies three basic properties of water-retention curves: a) matric potential is zero at saturation water content, b) matric potential approaches -∞ as water content approaches zero, and c) volumetric water content in dry soil is proportional to bulk density. Values of model parameters were taken from the literature, or estimated by fitting model predictions to data for sandy soils with low organic matter content. Most of the variation in water-release curves in the calibration data was explained by texture, with negligible effects of bulk density and sand particle size. The model predicted that variation in clay content among soils within the sand and loamy sand textural classes had substantial effects on water-retention curves. An understanding of how variation in texture among sandy soils contributes to matric potential is necessary for interpreting biological activity in arid environments.