Abstract: The pressure chamber of 600 ℃ and 20 MN servo-controlled triaxial rock testing system with high temperature and high pressure has adopted the piston-thick-walled cylinder structure. Compared with normal pressure chamber, the inner diameter rises to 300 mm, the outside diameter becomes 1 060 mm and the height grows to 1 100 mm. The size of test specimen also amounts to 200 mm400 mm. The calibration test applies internal heating type and electronic heating approach with low voltage and high current. Before the system is put into use, the reference temperature point should be calibrated. Thus, nine thermocouples were placed in and outside the specimen to observe the temperature variation at these observation points, and to study temperature field characteristics inside of specimen, as well as its relationship with reference point temperature. The calibration experiment results indicate that, the temperature data collected from the middle of the junction of NaCl and micanite paper are stable and reliable, and are in a favorable linear relationship with the inner temperature. Therefore, the place is qualified to be the reference temperature point in the heating process. The test results show that the inner temperatures of the specimen are higher than the temperatures of the junction of NaCl and micanite paper in the same cross section, the middle temperatures of whom are higher than those on both ends. In the inner axial direction, the upper end and the bottom temperature are symmetric to the middle of the specimen whose temperature is higher than either of those of the both ends. The disparity from middle to end grows bigger as the heating temperature increases, the maximum gradient is 0.85 ℃/mm. In the radial direction, the larger distance from the central point, the greater trend of linear temperature distribution. While the central temperature is the lowest, the surface temperature is the highest. The biggest temperature gradient is about 0.5 ℃/mm. The temperature at the central point, 40 mm from the central point, 80 mm from the central point and on the surface of the specimen are respectively larger than the reference temperature by 21%, 27%, 28% and 29%. Whereas, in the inner axial direction, the upper end temperature and the bottom temperature are close, only 1.5% lower than the reference temperature. It is testified that the heating approach is reasonable, and the accuracy of temperature control is high and the temperature of test specimen is well-distributed.