Improved interpretations of the strength of the San Andreas Fault near Parkfield, CA based on thermal data require quantification of processes causing significant scatter and uncertainty in existing heat flow data. These effects include topographic refraction, heat advection by topographically-driven groundwater flow, and uncertainty in thermal conductivity. Here, we re-evaluate the heat flow data in this area by correcting for full 3-D terrain effects. We then investigate the potential role of groundwater flow in redistributing fault-generated heat, using numerical models of coupled heat and fluid flow for a wide range of hydrologic scenarios. We find that a large degree of the scatter in the data can be accounted for by 3-D terrain effects, and that for plausible groundwater flow scenarios frictional heat generated along a strong fault is unlikely to be redistributed by topographically-driven groundwater flow in a manner consistent with the 3-D corrected data. Copyright 2004 by the American Geophysical Union.