An effective approach for short-term load forecasting by using hybrid extreme learning machines and multi-resolution analysis was proposed in this thesis. Two studied cases were provided to verify the effectiveness of the proposed approach. To further examine the performance, a comparison with existing methods including autoregressive integrated-moving average models, back propagation neural networks, and grey rolling models was given to highlight the merit of the proposed approach. The hourly load data for this experiment are derived from a realistic Taipower substation with four 250 MVA transformers during the summer period of 2008. The performance is evaluated by RMSE (root mean square error) and MAPE (mean absolute percentage error). Experimental results showed that the proposed approach is superior to existing methods.