Abstract: An extreme frozen ice and snow disaster brings severe losses in Anhui Province during early 2008. Some overhead transmission lines are destroyed seriously during this event. It is important to investigate the characteristics of ice accumulation and develop appropriate models to estimate the disaster, but the research doesn't go smoothly due to the lack of data. If a quantitative relationship can be built between the meteorological elements and the ice thickness by the wire icing observation data of fewer stations, the ice-covered mechanism can be understood better, which can also lay the foundation for disaster risk regionalization.Based on the long-term wire icing observation data and climate data of 16 meteorological stations in recent 50 years, several models are built by stepwise multiple linear regression and artificial neural networks, to calculate standard ice thickness of wire icing in Anhui Province.Through coordinated experiment, the optimal models built by stepwise multiple linear regression are confirmed which handle temperature, humidity, wind speed of the icing day, one day before and two days before. The models by artificial neural network perform better in simulating comparatively, but in the prediction they are very unstable, and less effective. The model driven by 26 meteorological factors (data from 1987 to 2008) perform better than the model driven by 24 factors (data of from 1960 to 2008). The best models are picked out by the results of simulating and predicting. And the absolute deviation is 1.1 mm to the north of the Huaihe River, 1.3 mm to the south of the Huaihe River, 6.6 mm in the area of high mountains, and the relative deviation rates of the three models are about 60%—70%. The result is better than other references. For the mechanism of icing, temperature, humidity and wind are key factors, among which temperature is the most influential one. The ice thickness of the plains and hilly areas are mainly affected by weather conditions of the day, while that of mountain areas are also affected by the weather conditions of a few days before. Finally, the ice-wire thickness at the meteorological stations without wire icing observation is calculated by the optimal model, which proves this method feasible.