The invention provides an online historical data tendency analysis method, and the method comprises the steps: A, analyzing the stability of an online historical data sequence; B, carrying out the stationary processing of the online historical data sequence through employing a difference method; C, determining p and q in an ARIMA(p, d, q) model; D, observing the normal distribution of the ARIMA model under an exponential smoothing model; E, carrying out the online analysis and prediction of online historical data. The method can achieve the analysis of the historical data of a national power system, and achieves the prediction of the future tendency of power network tide based on the ARIMA mode of time sequence through a series of data obtained according to the time sequence.