As the traditional gray forecasting model GM(1, 1) has poor performance in forecasting the fastgrowing power load, we present a chaotic co-evolutionary particle swarm optimization (CCPSO) algorithm, one with better e±ciency than the PSO algorithm. Based on the GM(1, 1) model, the CCPSO algorithm is adopted to solve the values of parameters a and b in GM(1, 1) model. This is how the way we come up with the CCPSO algorithm-based GM. As can be seen ¶ Corresponding author.from experimental results of case simulation on the power consumption in three regions, the CCPSO-GM model is better than the other four forecasting models, which attests its wide applicability and high forecasting accuracy.