This paper presents a study proposing a method to assess the condition of metal-oxide surge arresters. Thermal data using thermal imaging as well as the leakage current third harmonic component were used as tools to investigate the surge arrester aging condition. Artificial Neural Network was employed to classify surge arrester condition with the temperature profile, ambient temperature and humidity as inputs and third harmonic leakage current as target. The results indicated a strong relationship between the thermal profile and leakage current of the surge arrester. This finding suggests the viability of this method in condition monitoring of surge arrester.