In accordance with the complex internal cooling structure of large-scale power transformers and the problem on temperature prediction, we established a fluid network model based on the oil circuit structural characteristics and the analysis of winding cooling features. In this model, we divided the cooling circuit and calculated the local flow field by computational fluid dynamics (CFD),then analyzed the flow resistance and the cooling characteristics. The overall fluid network was built by coupling with the oil flow state, and the global temperature field distribution was determined. Under the condition of ensuring precision, the model size and the calculation time could be obviously reduced by using this method. An oil-immersed power transformer with a capacity of 240 MVA and a voltage class of 330 kV was taken as an example for calculation. The final results show that the hot point, whose temperature is 77.59 ℃, and relative temperature rise are 53.43 K to the bottom oil and 57.59 K to the air, is located at the top of the HV winding. In the end, the thermal imagery test and the optical fiber temperature test prove that the calculation of oil circuit temperature rise is accurate, and the difference between the theoretical calculation figures and the results of temperature test experiment using optical fiber is 3.57% which is enough for engineering design. As a result, the presented method can be used in temperature calculation of electrical equipment with other structures.