The power transformer is one of the most expensive components in an electrical transmission and distribution network. The temperature inside a power transformer varies with the load. The higher the operating temperature the faster the degradation of paperpress board insulation, consequently the remaining life of the transformer decreases. In order to keep the temperature under a safe limit, thermal models for predicting the hottest spot temperature in a transformer based on the load, cooling factors and ambient conditions have been developed and currently used in practice. In this project, a thermal model stipulated by the IEC standard is evaluated. The evaluation is carried out using the experimental results obtained at varying load current on a 40MVAOFAF-cooled transformer equipped with both fibre-optic sensors and oil temperature indicator, and a 63MVA-ONAF-cooled transformer equipped with oil temperature indicator. The model parameters are obtained from a normal heat-run test performed by the transformers manufacture. It can be shown that the response of the Difference equations method to the load variations is better than the Exponential method. A model Using Difference equations method and based on the internal cooling variations is proposed in this thesis. However, the calculated hot-spot temperature is not in a good agreement with the measured hot-spot temperature during rapid load increase.A proper result for the hot-spot temperature calculation using Difference equations method with ONAN cooling mode is obtained, although the cooling class of the transformer is OFAF. The calculated hot-spot temperature is compared to the measured hot-spot temperature to evaluate the model. The Difference equations method with ONAN cooling mode never underestimates the hot-spot temperature more than 3℃under a normal loading period. For the particular overloading situation the underestimation is 9.9% of the time period that is less accurate comparing to the normal loading period. The rate of occurrence of the underestimation is 7.7% for the Difference equations method with OFAF cooling mode during the normal loading situation and 13.8 % during the overloading situation. However, the Difference equations method with OFAF cooling mode never underestimates the hotspot temperature more than 10℃ during the normal loading and overloading situations. Improvements are still required in order to predict the hot-spot temperature accurately using the Difference equations method under various dynamic loading conditions. Results indicate that the calculated top-oil temperatures are either underestimated or overestimated. For accurate hot-spot temperature calculations, it is hence recommended to use the measured top oil temperature.