Abstract: There is a significant difference in the decadal variation between early summer (June) and mid-summer (July–August, JA) precipitation over the Haihe River basin; particularly after 2002, the decadal variation characteristics of June and JA precipitation over the Haihe River basin are opposite; thus, it is necessary to establish independent prediction models for June or JA precipitation. Based on advantage of annual increment approach, some important predictors (e.g., the previous winter monsoon, ENSO phase and its developing speed) associated with the annual increment of mid-summer rainfall anomaly over Haihe are chosen using correlation analysis. The difference between the current year and previous years (DY) of the sea level pressure index (SLPI) of key areas in middle and high latitudes of Europe and Asia in the previous winter, the Niño3 index in June, and the tendency between June and January of the Niño3 index as ENSO evolution speed are used as key factors to establish the multivariate linear regression equation. Then, the forecast experiment of mid-summer precipitation over the Haihe River basin in 2022 is performed according to the predicted Niño3 index in June by models. The comparison between the annual increment and climate models results started in March revealed that the former has a high prediction ability, particularly in flooding years. Furthermore, the failure hindcast case is carefully examined through the contribution of each predictor. The main factor is the SLPI, which discloses the relationship between East Asian winter and summer monsoon, which strongly relies on the following tropical Pacific and Indian Ocean sea surface temperature anomalies (SSTa) evolution. Nevertheless, the tropical SSTa, particularly in the western Indian Ocean in June, shows unique features and may disturb the contribution of the SLPI. This implies that the near prediction of SSTa in crucial areas highly related to the Haihe mid-summer rainfall should be paid attention to.