This paper presents the study of Support Vector Regression (SVR) to forecast the future streamflow discharge using past streamflow and rainfall data, which is closely related to regularization network and Gaussian processes. A Gaussian Radial Basis Function (RBF) kernel framework was built on the dataset to optimize the tuning parameters and to obtain the moderated output. It has been observed from various studies that the prediction ability of RBF kernel function is better for regression problems. The training process of Support Vector Machine (SVM) involves the selection of both kernel parameters and regularization constants. The constants such as $\gamma$, $\epsilon$, C and $\sigma$ are the parameters of SVR which were optimized to obtain the desired outputs. Where parameter C determines trade-off between model complexity and degree to which deviations are larger than $\epsilon$ are tolerated in optimization formulation and the parameter $\epsilon$ controls the width of $\epsilon$-sensitive zone. The study area include the upstream part of Jayakwadi dam. The prediction is based on flow at two Gauge and Discharge Sites (GDS) i.e. Kopargaon and Nagamthan and rainfall data at three Standard Rain gauge Stations (SRG) i.e. Rahata, Khirdisathe and Wadala Mahadev in Maharashtra, India. The daily data values of these stations for 18 years were collected from Hydrological Data User Group (HDUG), Nashik. About 70% of data were used for training purpose and remaining 30% of data were used for testing purpose. The experimental results showed that the SVR algorithm is reliable and efficient method for streamflow prediction, which has an important impact on the water resources management of region.