Energy conversion device suffer from thermal loading as a result of inefficiencies during their operation which may lead to device degradation and possible failure. It is of interest to monitor the internal temperature of the device to ensure its safe operation. Mathematical models of different complexities have been developed for the purpose of real time temperature monitoring. Temperature estimation accuracy is dependent on the thermal parameters such as the material conductivities and convective heat transfer coefficients. The complex construction of such devices means that the exact value of the thermal parameters is often not known. This paper presents the use of an inverse identification technique for the estimation of thermal parameters of an axial flux permanent magnet device designed for vehicular applications. The proposed method provides a practical approach to determine the thermal parameters indirectly from temperature measurements. A constraint least square method coupled to an analytical solution of the one step ahead predictor of temperature is used for parameter estimation. A parametric study is performed and it is shown that some of these parameters vary as a function of the operating point of the device. The estimated parameters are then used in an analytical thermal model for real time temperature monitoring during a drive cycle. A maximum time averaged error of 1.8°C or an equivalent error of about 3% of the measurement range is observed for the estimated winding temperature.