The precise and accurate models of hydrological time series that are embedded with high complexity, nonstationarity, and non-linearity in both spatial and temporal scales can provide important information for decision-making in water resources management and environmental related issues. Hybrid wavelet transform (WT) and adaptive neuro-fuzzy inference system (ANFIS) has been used in this study to improve the forecasting capability of ANFIS model by decomposing the time series into sub-time series (approximation and details) using wavelet transform then combining the effective and significant time lags of sub-time series to form a set of input variables. The present study attempts to add the effective and significant time lags of original time series as extra variables to the input variables set. In addition, different combinations of variables, 1-3, from the set of input variables as inputs to the ANFIS model were used to forecast the time series.To examine the potential of the approach for practical applications, the model is applied to forecast, one step-ahead, the monthly data of hydrological time series (rainfall, evaporation, minimum and maximum temperature, average wind speed and reservoir inflow) for Kirkuk, Sulaimani, Dokan and Darbandikhan meteorological stations in Iraq. The best fit models were selected using the coefficient of determination ( ) and root mean square error ( ). Based on the results, the proposed model has high performance in forecasting the monthly minimum and maximum temperature, evaporation and reservoir inflow with values ranged from 0.93 to 0.99 and relatively good performances in forecasting the monthly rainfall and average wind speed with values ranged from 0.77 to 0.93.