Management of surface water resources systems in water shortage conditions has many problems due to many uncertainties, unexpected river flows, and the multipurpose nature of water supply systems. One of the policies for operating reservoirs in drought conditions is the hedging rule method. In this method, threshold limits are considered for water demands. If the supply of water is subject to stress, the amount of supply is limited using threshold limits, and water is distributed in a way that water is stored for water shortage periods. One of the problems that arise in reservoir real time operation is forecasting the inflow into the reservoir in future periods. This leads to uncertainties arising from the prediction. In this research, the hedging rule method in the operation of the reservoir is used, for a number of future time steps (decision horizon). The forecast horizon is determined in order to decide the number of future time steps that affect the decision making in decision horizon. The application of this method is expressed in avoiding additional uncertainty in the operation model. Considering that the share of agricultural section demands is significant relative to the demands of the drinking, industry and the environment sections, so in another section of this research, the economic model was developed with the aim of maximizing the benefit from agricultural crops and it was optimized using the genetic algorithm. Afterwards, to investigate the effect of the hedging rule model constraints on the results of the economic model, a combined model of economic-hedging rule was developed in the form of the two-objective model and was optimized with NSGA-II algorithm. The results of the two-objective model were assessed with the TOPSIS method and were compared with the results of the economic model. The results of the implementation of the hedging rule policies in the case study of Karkheh Dam indicate that although the reliability index for supplying water demand was reduced from 40.1% in observation to 34% in the hedging rule model, but the severity of the vulnerabilities decreased from 100.3 (mcm) in observation to 48.79 (mcm) in the hedging model. The results of forecast horizon implementation showed that in order to make a decision for the first time step in the future, only the data of the four subsequent periods will affect the results of the first step. Considering the economic model, it is shown that, in the current situation, with the cultivation of 135991 (ha) of crops, the obtained benefit of 30 thousand dollars, which has increased by 35 thousand dollars (12%) in the economic model without development of the cultivated area (134991.4 (ha)) and with cultivation of crops with higher yields. Also the gross consumption of water in agricultural land per hectare (10362.2 (m^3/ha)) has decreased (823 (m^3/ha)) relative to the observed condition (11185.4 (m^3/ha)). By Regarding the entry of hedging rule constraints in the two-objective problem, the results showed that, the total cultivated area in the two-objective model has decreased by 11% compared to the economic model. The reliability index in supplying water demands in the two-objective model (5.3%) compared to the economic model (38.6%) and the severity of the vulnerability in the two-objective model (44.3 mcm) compared to the economic model (84.9 mcm) has decreased. Therefore, in two-objective model, by accepting more shortage periods (reduced reliability), severe water shortages and failure of the system have been prevented (reduced vulnerability). Also, by determining the reservoir rule curve and applying the hedging coefficients as decision variables, the release from the reservoir has decreased by 12% compared to the economic model. This has led to an increase in the volume of water stored in the reservoir compared to the economic model and as a result, the hydropower energy produced in the two-objective model (7.5*106 MWh) has increased compared to the economic model (7.4 *10 6 MWh). Keywords: hedging rules; forecast and decision horizon; economic-hedging optimization; NSGA-II algorithm; TOPSIS method.