Hydraulic gradients in near-surface aquifers in Memphis, Tennessee, USA, have reversed from upward to downward as a result of municipal and industrial pumping over the past 100 years. This change in hydraulic conditions has created potential for downward migration of surface and shallow-aquifer water through the upper Claiborne confining unit to the Memphis aquifer, the main water source in the Memphis area. Previous studies have identified regions where the confining unit is thin or absent and have demonstrated downward leakage through several discreet windows. In the present study at the Sheahan municipal well field in Memphis, water chemistry, ³H/³He, geophysical, and geologic data were used to clarify groundwater leakage processes and mixing fractions of leakage and native Memphis aquifer water. The tritium and geochemical data were interpreted to reflect mixing and reactions between two end-member water compositions: a modern Na-SO₄-Cl-rich end member from leakage and a submodern Ca-Mg-HCO₃-rich end member from deep groundwater in the Memphis aquifer. The ³H and ³He data from five wells resulted in apparent ages ranging from 16 to 51 years that decrease with depth. These samples were interpreted to be a binary mixture of relatively young modern water and relatively old submodern water. The apparent ages of the two shallowest samples were coupled to the age of the youngest water and, thus, reflect the average mass time of modern recharge. The ³H/³He data were interpreted to indicate that the upper part of the Memphis aquifer is being recharged with modern water that left atmospheric contact less than 20 years ago. The geologic data indicate that hydrogeologic pathways, such as heterogeneities in the upper Claiborne confining unit and paleovalleys incised into the upper Claiborne confining unit, exist for vertical flow of leakage from the surface to the Memphis aquifer. The most reasonable water source for leakage in much of the Sheahan well field is a nearby stream, Nonconnah Creek. Mixing ratios of shallow aquifer water near Nonconnah Creek and deep Memphis aquifer water were evaluated using end-member water chemistry data, a series of probable geochemical reactions, and a mass-balance computer code (NETPATH). The best set of model results assumed that some of the sodium and chloride in the mixture were extracted from brackish or marine formation waters in the confining unit. In this case, the waters containing the highest total tritiogenic ³He+³H reflected mixtures of 22 to 32% shallow aquifer water with deep Memphis aquifer groundwater. Accompanying reactions include dissolution of dolomite, cation exchange reactions, dissolution of CO₂, precipitation of calcite and variable reactions involving Na-montmorillinite, goethite, and pyrite. The identification of a prominent leakage pathway into the upper part of the Memphis aquifer raises concerns about the source and quality of the water leaking into the upper part of the aquifer. Furthermore, the results clarify the need to evaluate connections between surface water and groundwater to develop meaningful groundwater protection strategies.