Background and Objective: Selecting appropriate inputs for intelligent models are important because it reduces the cost and saves time and increases accuracy and efficiency of its models. The aim of the present study is the use of Shannon entropy to select the optimum combination of input variables in the simulation of monthly flow by meteorological parameters. Method: In this study, meteorological data and monthly time series of discharge of Zarrinrood River (Safavankeh Station) in East Azarbaijan from 1336 to 2015 were used. The meteorological parameters and the month of the year were considered as inputs in the entropy method to determine the effective composition. Results: Shannon entropy results showed that the rainfall parameters, month of year and temperature provide better results for modeling. The simulations were performed using intelligent hybrid models of particle swarm hybrid algorithm and hybrid simulation hybrid algorithm. Discussion and Conclusion: The results showed that among these models with the same input structure, the hybrid algorithm simulation based on the support vector machine had better performance for simulating the flow rate compared to other intelligent hybrid models. The results also show that the entropy method is good for selecting the best input combination in smart models.