SUMMARYEffective prediction of the contamination level of insulators is an important approach to the prevention of pollution flashover accidents, and the leakage current (LC) is an efficient factor to analyze and detect the contamination level of insulators. In order to predict how severe the contamination level of insulators is and prevent pollution flashover accidents, firstly, some artificial pollution tests have been investigated under different contamination levels, relative humidity (RH), and ambient temperature (T). Secondly, based on the experimental data, the three characteristics of LC, namely the maximum pulse amplitude (I h ), the energy ratio (K), and the energy (E), have been extracted. They reflect jointly the LC from different perspectives. Thirdly, the variation laws between the three characteristics and the contamination level, RH and T, have been obtained. Fourthly, the prediction of the contamination level by using least squares support vector machine (LS-SVM) model has been presented, in which the three characteristics, RH and T are used as the inputs of model, and the contamination level is used as the output of model. Finally, this model has been verified with the test data obtained in the artificial climate chamber, and the prediction results are consistent with the test results. Therefore, the LS-SVM model is acceptable to predict the contamination level of insulators and is of significance for the prevention of pollution flashover accidents.