The aim of this paper is to investigate the teleconnection of large-scale climatic signals with river flow discharge in northern Iran. To this end, the Garmrud, Alishrud, and Babolrud rivers were selected: their monthly data set during the years 1993–2018 was received from the regional water company of the Mazandaran Province. Climate indices including the Southern Oscillation Index, extreme eastern tropical Pacific sea surface temperature (Nin˜o 1 ? 2), Global Mean Temperature Index, North Pacific pattern (NP), Pacific Decadal Oscillation, and North Atlantic Oscillation were investigated. The results of the crosscorrelation, and the Spearman’s test showed that there is a significant relationship at a 0.01 level between the indices and the river flow discharge. The most correlated indices were the NP and Nin˜o 1 ? 2. The NP index was inversely correlated and the Nin˜o 1 ? 2 was directly correlated with the rivers’ flow variations, and the time lag of these two indices’ maximum effectiveness was 2 months and 1 month, respectively. The adaptive neuro-fuzzy inference system and multilayer perceptron (MLP) models were used as the predictor models. The inputs were selected from the time lags of both river discharge and climatic indices. According to the normalized root mean square error (NRMSE) values (0.14–0.22), the predictions are evaluated as having moderate and appropriate qualities. The best performance of all models, R = 0.824 and RMSE = 0.350 CMS, was achieved for the MLP model. The results showed that consideration of the climatic indices as inputs may improve monthly flow predictability by an average of 24.1% (11.4–38.4% among the rivers and models).
