Ten thermocouple psychrometers (TCPs) to measure water potential (WP) were installed in three holes in G-Tunnel at the Nevada Test Site as part of the Prototype Engineered Barrier System Field Tests. These integrated tests measured several parameters as a function of location and time within a few meters of a heater emplaced in welded tuff. The primary goal of the TCP experiment was to find out whether the combination of laboratory calibration and field use of the TCP can provide useful data for determining the change of moisture condition in the field. We calibrated the TCPs in NaCl solutions up to 80°C (176 G F) in the laboratory. In two holes, we used rubber sleeves and packers to house TCPs, and in the third hole, we used foam. All three holes were grouted behind the TCP assemblages. Field results of the heater test showed that small temperature gradients were present for all measurements. Nevertheless, the WP calibration made die necessary correction for the nonisothermal condition. The initial moisture condition indicated by TCP data was about 99.5% relative humidity or a WP of about -5 bar. This corresponded to 15.4 g/m 3 of water in the air near the borehole wall, which was much wetter than we expected. A drying and re-wetting cycle peaked at about day 140 with a WP of-65 bar in borehole P3, located below the heater. A similar cycle but reduced in scale was found at about day 175 with a WP of-45 bar in borehole P2, above the heater. Tnis difference in drying behavior above and below the heater was also observed from neutron data and was explained as a gravity effect. As temperatures increased, the evaporation rate of pore water increased. In unfractured rock, the gas-phase flow was primarily outwej-d. Water condensed above the heater would drain back to keep the boiling region wet, but water condensed below the heater would drain away from die boiling region. This conceptual model explained both the time and magnitude differences for data from holes above and below the heater.