Xin’anjiang model was used to simulate the flood process of Qujiang River watershed of Qiantang River. The effectiveness of six real-time correction methods for correcting flood forecasting results in the study area was evaluated based on indexes such as Nash efficiency coefficient (NSE), relative error of flood peak (RE), and peak time error (Δ T ). These methods included real-time correction method, real-time correction method by feedback simulation, autoregressive (AR) method, random forest (RF), k -nearest neighbor algorithm (KNN), and artificial neural network (ANN). The results show that all six correction methods can reduce RE, with RF being the best, followed by the real-time correction method and the method by feedback simulation. In terms of NSE, ANN and AR methods perform well, especially when the starting forecast time is far from the flood peak; ANN shows a better performance. In terms of peak time, RF has the best correction performance, followed by ANN. Overall, ANN shows the best performance, which can improve the accuracy of flood forecasting to a certain extent.