This paper presents a framework for studying regional water balance in which the physical processes are first described at the local instantaneous scale and then integrated to the annual, basin-wide scale. The integration treats the relative soil saturation (i.e., the soil moisture divided by the porosity) and precipitation intensities as stochastic variables in space and time. A statistical equilibrium characterizes the annual water balance, resulting in a specific relation that predicts the space-time average of soil saturation in terms of soil, climate, and vegetation parameters. Specific relationships are proposed to relate the space-time average soil saturation to runoff, groundwater recharge, and evapotranspiration. This framework is applied to the Illinois River Basin. The shape of the spatial and temporal distributions of soil saturation are determined from observations. The other parameters are determined from the physical characteristics of the basin and calibration procedures. The resulting model is able to reproduce an observed relation between the space-time average soil saturation and precipitation. It is also able to reproduce observed relations between space-time average soil saturation and space-time average evapotranspiration, surface runoff, and groundwater runoff.