The invention discloses a method for judging hydrological time series trend types. The method comprises the following steps: evaluating the significance of the linear trend of a sequence TS (t) by using a generalized least square method; judging the trend type of the sequence TS (t) by utilizing the determined KPSS and PP inspection methods; if the results of the two methods are consistent, givinga trend type result; selecting a Pettitt method to identify a mutation component B0 of the sequence TS (t) if the two are contradictory, and carrying out discrete wavelet decomposition on the sequence TS (t); removing a low-frequency component L (t) and a mutation component B0 of the sequence TS (t), taking the remaining components as a reconstructed sequence TS '(t), and identifying the trend type of the sequence TS' (t) by utilizing the determined KPSS and PP inspection methods; if the results of the two methods are consistent, giving a trend type result; analyzing an autocorrelation coefficient graph and a partial correlation coefficient graph of the reconstructed sequence TS '(t) if the two are still contradictory, judging as a deterministic trend if the two show truncation characteristics, judging as a random trend if the two show trailing characteristics, and finally judging that the trend type of the sequence is unknown if the results of the two are contradictory.