In this paper, a method is presented in which the Markov correction-based load prediction model is applied to the power efficiency monitoring terminal. First, the historical load data is pre-processed by use of an exponential smoothing method, and the relative error is computed; Second, the state transition matrix of the relative error obtained by an exponential smoothing method is computed by use of the Markov correction-based load prediction model, so that the prediction interval of power load in the target days is acquired; Finally, an example is taken to compute the relative error of the Markov correction-based load prediction model, yield a more accurate range of the prediction values, and apply this method to power efficiency monitoring terminal accurately. The experimental results show that in light of the prediction interval of power load in the target days, the Markov correction-based load prediction model has a higher precision than that of the exponential smoothing method. Moreover, the predicted results are more credible with aforementioned method.