Inventory-prediction equations describing snowpack water content as functions of readily available or easily obtained inventory variables were developed for use in the ponderosa pine type in Arizona. Although empirical in nature, these equations include parameters assumed to index interception of precipitation inputs, obstruction of direct beam solar radiation, and re-radiation from trees onto the snowpack. Primary consideration was given to forest cover variables in synthesizing the inventory-prediction equations I because currently proposed water improvement programs designed to increase water yield derived from snow consist essentially of vegetative manipulations. Additional independent variables evaluated include potential direct beam solar radiation, elevation, soil, and precipitation inputs. All of the inventory-prediction equations describing a particular snowpack condition were not statistically equivalent in terms of the standard error of estimate or the coefficient of determination. Equations including basal area, bole area I volume, and height-index as expressions of forest cover density were generally better than equations with point density, sum of diameters, and number of trees. Inventory-prediction equations developed to describe snowpack dynamics throughout the accumulation period showed similar statistical form, except as possibly attributable to different precipitation inputs. Equations for characterizing residual snowpacks during spring runoff were statistically weak, possibly because factors other than those considered in this study control the runoff process. The inventory-prediction equations were developed to estimate the mean snowpack water content on a basin, and to describe the trade-off , or the rate of exchange, between snowpack water content and forest-site variables on a decision-making unit. The equations do not necessarily predict changes in recoverable water yield resulting from the implementation of a land management system, however. Nonbiotic characteristics of the land, L e., topographic features, geologic formations, and soil . properties, could conceivably control water yield to the extent that changes predicted by the inventory-prediction equations could be masked. Because of limitations in predicting potential changes in recoverable water yield, it was assumed that a land management system that maximizes snowpack water content on site would also provide the maximum potential for increasing recoverable water yield derived from snow. Management guidelines designed to allow snowpack water content to be maximized on site can be formulated within the framework of the inventory-prediction equations, multiple use management constraints, and forest-based product benefits and costs. Management guidelines indicate that the greatest gain in snowpack water content on site would be realized on decision-making units where the greatest reduction in forest cover density could be prescribed. However, a timber production constraint may limit the array of management possibilities. This constraint was defined as 35 to 40 square feet of basal area or 1,050 to 1,175 cubic feet of volume per acre, depending upon the existing growth percent and the intermingling of tree volumes and size classes. The potential increase in snowpack water content on site will be determined by the magnitude of the reduction in forest cover density and how close management re-direction can approach the timber production constraint. The proportion of the snowpack water content on site converted to recoverable water yield is dependent upon the runoff efficiency.