Flood events can cause devastation in heavily inhabited areas along rivers; thus, obtaining longer forecasts with superior accuracy is vital to protect public and assets. A soft computing approach is presented for the prediction of floods in the Mahanadi River basin, Odisha. River Mahanadi is a major eastward flowing river in the Indian Peninsula that originates in Chhattisgarh before entering Odisha. During the monsoon season, many regions where the river flows through in Odisha experience severe floods. This makes flood forecasting essential for mitigation and proper management of the watershed. For this purpose, a Support Vector Machine (SVM) model and a metaheuristic algorithm known as Harris Hawks Optimization has been incorporated. 20 years of statistical data was utilized to train and test the model. The model’s reliability was evaluated using Root Mean Square Error (RMSE), Willmott Index (WI), and Coefficient of Determination (R2) as performance indicators. It was concluded from results that the proposed hybrid model produced superior results when compared against the conventional model.