Ground-water conductivity data were obtained from shallow wells in a 12 km2 stream-basin along a 400 m transect, extending from the divide to the stream. The stream, Pringle Branch, is a second-order perennial stream in Hillsborough County, Florida. The shallow stratigraphy consists of 2-3 m of fine sand over a layer of clayey silt and silty clay. Vegetation cover includes grasses on the upper and middle slope, and riparian woodlands on the foot slope and floodplain. Precipitation is about 1.3 m per year. Shallow ground-water conductivity is about 50 uS/cm at the divide. It increases moderately along the mid slope, then increases markedly within the riparian woodlands, reaching a maximum of about 500 uS/cm at 30m from the stream and then decreases to about 150 uS/cm at the stream. The spatial variation of terrain electrical conductivity data collected using electromagnetic methods (EM 31) is similar to the spatial variation of ground-water conductivity.Dry season through wet season monitoring shows that ground-water conductivity in each well varies about 40%, generally following variations in potential evapotranspiration (ETpan). The more than five-fold increase in ground-water conductivity from divide to riparian woodlands is maintained during both dry and wet seasons. The ground-water conductivity in this basin appears to be determined principally by spatial variations in ET and not by temporal variations in ET or interaction with soil minerals. The data suggest that patterns of ground-water conductivity can be used to infer patterns of ET variation within a small basin. A mass transport model constructed to test the hypothesis that evapotranspiration has the dominant effect on ground-water conductivity closely duplicates the observed variation in ground-water conductivity from divide to stream. The model uses two evaporation rates, 0.73 m/y for the grasses and 1.46 m/y for the riparian woodlands, and no contribution from solution of matrix materials.