The Modeling and prediction by time series analysis has been an important role in last decades and has wide applications in different fields of science and engineering, such as hydrology. So that parameter estimation of various time series models is one of the essential step of modeling. The presented methods in this field such as method of moments, have complicated formula in spatial and temporal multivariate models, particulary, so this need a lot of time and survey of different formula for each model. In this study, by using Adaptive Neuro-Based Fuzzy Inference Systems(ANFIS), a new and effective method for parameter estimation of various univariate and multivariate time series models is presented. Performance of this technique has surveyed by hydrologic data of the Zayanedrood river basin and then parameters of the models have estimated. After this, prediction of the time series has done by using these models. In addition, prediction of time series has done by ANFIS and MultiLayer Perceptron(MLP), too; and then the results have compared. For comparison of results, Mean Absolute Error(MAE) criterion is used. Results of simulations show that presented method in this study can be used as an intelligent and effective technique for univariate and multivariate time series modeling. Keywords: Hydrological Time Series, Prediction, Univariate, Multivariate, Adaptive Neuro-Based Fuzzy Inference System(ANFIS), Neural Networks.
