Steam-enhanced remediation (SER) has been successfully used to remove DNAPL and LNAPL contaminants in porous media. Between August and November 2002, SER was tested in fractured limestone at the former Loring Air Force Base, in Maine, USA. During the SER investigation, the U.S. Geological Survey conducted a series of borehole radar surveys to evaluate the effectiveness of radar methods for monitoring the movement of steam and heat through the fractured limestone. The data were collected before steam injection, 10 days after the beginning of injection, and at the end of injection. In this paper, reflection-mode borehole radar data from wells JBW-7816 and JBW-7817A are presented and discussed. Theoretical modeling was performed to predict the variation of fracture reflectivity owed to heating, to show displacement of water and to assess the effect of SER at the site. Analysis of the radar profile data indicates some variations resulting from heating (increase of continuity of reflectors, attenuation of deeper reflections) but no substantial variation of traveltimes. Spectral content analysis of several individual reflections surrounding the boreholes was used to investigate the replacement of water by steam in the fractures. Observed decrease in radar reflectivity was too small to be explained by a replacement of water by steam, except for two high-amplitude reflectors, which disappeared near the end of the injection; moreover, no change of polarity, consistent with steam replacing water, was observed. The decrease of amplitude was greater for reflectors near well JBW-7817A and is explained by a greater heating around this well.