In the covered karst of west-central Florida, USA, sinkholes (sandy collapse conduits) provide locally concentrated recharge to underlying aquifers. For water management, it would be beneficial to understand the rates at which collapse conduits recharge an underlying aquifer. Self-potential (SP) monitoring has promise as a noninvasive, relatively low-cost method for assessing temporal variability in flow. Previous studies suggested that negative SP anomalies over collapse conduits correspond to downward-flowing groundwater; however, before SP surveys can be reliable indicators of conduit flow, SP from ET, soil conductivity changes, redox, electrode effects, and cultural noise must be better understood. A year of continuous SP monitoring with two grids of approximately 15 Pb/PbCl/KCl nonpolarizing electrodes each was combined with high-resolution ground-penetrating radar surveys and intermittent water table monitoring over two small covered-karst conduits in Tampa, Florida, USA. Although variations in SP resulting from changes in cultural noise, soil conductivity, ET, redox, and rainfall were evident, separate and unrelated positive and negative SP anomalies episodically manifested over conduits, which suggested that conduit flow could be dynamic, not static. Three flow regimes in conduits were postulated: conduit permeability higher than in surrounding surficial sediments, conduit permeability lower than in surrounding surface sediments, conduit permeability matched surrounding surface sediments. Numerical steady-state 2D simulations in Comsol created the three postulated flow regimes and revealed that a different SP polarity could result from different rates of flow: positive SP corresponded to higher permeability conduits, negative SP corresponded to lower permeability conduits, no or minimal SP appeared when conduits and surrounding sediments had equal permeability. In these models, downward flow was not responsible for generating negative SP. To assess the hydrology of a conduit, it appears that SP should be monitored continuously. Further monitoring of field sites with hydrologic sensors is needed.