Thermal fatigue cracking may initiate at a T-junction pipe where high and low temperature fluids flow in from different directions and mix. Thermal stress is caused by a temperature gradient in a structure and its variation. The accurate simulation of the temperature distribution in structures is, therefore, important for estimating thermal fatigue. In this study, an experimental method using a T-junction pipe with thermocouples was developed. Wall temperatures in the experiment have to be obtained at the inner surface of the pipe to validate numerical simulation results. The difference of position between the inner surface and the measurement point where using thermocouples may, however, have an effect on temperature data. The numerical simulation results showed that the amplitude of temperature fluctuations was reduced to 54% and the difference of phase was 0.91 radians when the sinusoidal temperature fluctuation of 5Hz was applied from the inner surface to the thermocouple measurement points. These results showed that wall temperatures at the inner surface should be estimated from measured data obtained by the thermocouples. A transfer function was, therefore, obtained to calculate wall temperatures at the inner surface from measured data. In addition, the numerical simulation results showed that the amplitude and the phase of temperature fluctuations differed depending on existence of voids around thermocouples. Respective differences of it in the amplitude and the phase were about 5% and 3% when the sinusoidal temperature fluctuation of 5Hz was applied at the inner surface. These results showed that thermocouples should be installed in pipes without voids to measure accurate temperature fluctuations. A mock-up test showed that voids stayed behind the thermocouples when the thermocouples were brazed into the pipe wall at atmospheric pressure, but the voids disappeared when thermocouples were brazed in a vacuum atmosphere into the inner surface of the pipe.