A new method of Johnson noise thermometry was developed and experimentally evaluated. The absolute temperature Ts was determined by measuring the available thermal noise power, i.e., the product of the open‐circuit thermal noise voltage and the short‐circuit thermal noise current generated by a sensing resistor at temperature Ts. The measured thermal noise power from the thermometer is a linear function of absolute temperature. This new method is independent of the sensing‐resistor composition, the mass and nature of the charge carriers, and, in principle, the ohmic value of the resistor. Consequently, the method has a wide range of application, including use in the realization of an absolute thermodynamic temperature scale and especially in the measurement of high temperatures, particularly in nuclear reactors where the properties of sensor materials not only change due to the aging effects induced by the high temperatures but also change drastically because of radiation damage and transmutation. An evaluation of the method was performed with the sensor and signal processor both located in an area of high electrical background noise. The signal cable connecting the sensor to the signal processor was about 0.5 m in length. The deviations of the experimental data were not more than ± 0.10% from a straight line calibration through absolute 0 K in the temperature range ΔT from 725 to 1275 K. In the range ΔT, the maximum error in the indicated temperature for a 100% increase in the value of the sensing resistor at a fixed temperature was ± 0.60% and for a 50% decrease, was −1.00%. The feasibility of separating the thermal noise signal from electrical background noise induced in a long signal cable was demonstrated with the sensor and signal processor connected with a 30 m long coaxial cable. With the extraneous noise power in the cable, more than an order‐of‐magnitude greater than the thermal noise power, the uncertainty in temperature due to the extraneous noise was less than 1% at 100 C.