The Bouwer and Rice method of estimating the saturated hydraulic conductivity (Ks) from slug-test data was evaluated for geometries typical of hand-dug wells. A two-dimensional, radially symmetric and variably saturated, ground water transport model was used to simulate well recovery given a range of well and aquifer geometries and unsaturated soil properties, the latter in terms of the van Genuchten parameters. The standard Bouwer and Rice method, when applied to the modeled recharge rates, underestimated Ks by factors ranging from 1.3 to 5.6, depending on the well geometry and the soil type. The Bouwer and Rice analytical solution was modified to better explain the recovery rates as predicted by the numerical model, which revealed a significant dependence on the unsaturated soil for the shallow and wide geometries that are typical of traditional wells. The modification introduces a new parameter to the Bouwer and Rice analysis that is a measure of soil capillarity which improves the accuracy of Ks estimates by tenfold for the geometries tested.