Arid-site recharge, while generally low, can be highly variable. Recharge under similar climate and soil conditions but with different plant cover and topography can vary from zero to more than the annual precipitation. Simple estimates of recharge based on fixed fractions of annual precipitation are misleading because they do not reflect the plant and soil factors controlling recharge. Detailed water balance models, successful for irrigated agriculture, fail to predict evapotranspiration accurately under conditions where plants suffer seasonal water stress and cover is sparse. Recharge, when estimated as a residual in water balance models, may be in error by as much as an order of magnitude. Similar errors can occur when soil water flow models are used with measured or estimated soil hydraulic conductivities and tension gradients. Lysimetry and tracer tests offer the best hope for evaluating recharge at arid sites, particularly in siting waste disposal facilities, where reliable recharge estimates are needed. Quantification of drainage using lysimetry over several years under a given set of soil, plant, and climate conditions for a specific site can provide a basis for calibrating models for recharge prediction. Tracer tests using such long-lived tracers as 36Cl or perhaps stable isotopes (180, deuterium) can provide qualitative estimates of recent recharge at a given site.