[1] Surface heat flow in the California Coast Ranges near Parkfield, California, exhibits substantial scatter, with differences as large as 20 mW m −2 over lateral distances of 5-70 km. This scatter has been an important limitation in using the heat flow data set to constrain geodynamic processes, but to date it has not been explained. Here we use a numerical model of coupled fluid and heat transport to test the hypothesis that heat advection by groundwater flow can generate the scatter. Our study significantly extends previous investigations in that we consider realistic and heterogeneous permeability architecture and topographic driving forces. We find that the magnitude and spatial characteristics of the scatter, including the standard deviation, variation in heat flow as a function of separation distance, and patterns of heat flow and elevation can be generated if the Tertiary sediments in the upper 2-3 km of the crust have a permeability ≥3 × 10 −16 m 2 , allowing recharge of ∼0.5 cm yr −1 or higher. These permeabilities and recharge rates are consistent with existing constraints on both quantities, suggesting that groundwater flow offers a plausible explanation for the observed scatter in the heat flow data set. Last, although not the primary focus of this study, we demonstrate that for a range of reasonable permeability architectures, topographically driven groundwater flow would not mask a thermal anomaly associated with frictional heating on the San Andreas Fault.