Monthly streamflow forecasting plays an important role in long-lead water resources planning and management. In the current paper, model fusion technique has been used in order to increase the accuracy of monthly streamflow forecast of Karkheh River at the entrance of Karkheh reservoir in winter. For this purpose, five models including: Artificial Neural Network (ANN), Generalized Regression Neural Network (GRNN), Support Vector Regression (SVR), K-Nearest Neighbor (KNN), and Linear Regression (LR) with optimized structure have been applied as individual forecasting models (IFMs). In order to combine the IFM models, two model fusion strategies including constant and variable weighting based on ordered weighted averaging (OWA) have been performed, where the Orlike method has been applied to determine the weights of IFMs. The results show that variables weighting strategy is more performable than constant weighting strategy in order to promote the accuracy of the forecast results. In addition, the comparison of the two strategies with two strategies including model fusion with artificial neural network and selecting the best IFM reveals that variable weighting strategy can significantly promote the accuracy of the forecast results than the latest strategies; such that this strategy increases the accuracy of the results 51.8, 38.1, and 44.5 percent as compared to ANN model fusion, and 7.6, 132, and 52.9 percent as compared to the best IFM for January, February, and March, respectively.