Heat flow was simulated to estimate riverbed hydraulic conductivity at three sites of induced infiltration adjacent to the Great Miami River in southwest Ohio. Three approaches were compared: 1D steady-state, 1D, 2D and 3D transient simulations. Preliminary results indicated that where strong downward gradients existed, 1D and 2D transient modeling performed equally well. The 1D transient model did not perform as well at sites with strong horizontal flow and weak downward gradients. The 1D transient approach produced results that showed similar trends to the 2D transient models in predicting short-term temporal changes of riverbed conductivity, but the absolute estimates were sometimes dissimilar. The 1D steady-state approach sometimes produced very similar results to the 2D transient models, but at other times it over-predicted riverbed conductivity. While the reasons for these observations are still under investigation, it is clear that 2D models are more appropriate in many circumstances for predicting hydraulic conductivity at sites of induced filtration. The 3D approach is expected to produce more accurate results and is still under investigation.