Analysis of electro‐thermal behavior in various operating condition based on finite element method was proposed in this article. Power frequency and impulse stresses vary the operating region and temperature of surge arresters. Therefore, adaptive power loss was considered as a source in the modeling of surge arrester based on finite element method. Adaptive network based fuzzy inference system and frequency dependent model were used to model power loss in low and high current regions, respectively. To obtain accurate power loss, which is very important factor in electro‐thermal analysis, applied voltage and temperature were considered as inputs in adaptive network based fuzzy inference system model. The experimental setup including impulse generator, power frequency high voltage generator and also an oven was arranged to achieve real data. Finally, the results of simulation based on proposed adaptive method were compared with the thermal infrared camera results. This study shows that the electro‐thermal behavior of surge arrester is predictable with a good accuracy by using artificial model. © 2015 Wiley Periodicals, Inc. Comput Appl Eng Educ 23:910–920, 2015; View this article online at http://wileyonlinelibrary.com/journal/cae; DOI