Meiyu precipitation of Zhejiang province was abnormally less than usual in 2018.During the Meiyu period， mean feature of large-scale circulation in mid-high latitude on 500 hPa exhibited a “trough-ridge-trough” pattern， while in low latitude， western Pacific subtropical High （WPSH） marched eastward and northward.Moreover， the upper-layer westerly jet over East Asia located northward and low-level jet supply south of Zhejiang was not sufficient.All the factors above worked together to form weak Meiyu.Inadequate moisture transport in the first and second ten days of June， inferior convergence of north and south flows， southward-placed westerly jet over East Asia， stronger South China Sea summer monsoon （SCSSM）， and weaker Indian monsoon acted jointly to cause a late Zhejiang Meiyu.There were three rainfall processes （process I， process II， and process III） and a time with few precipitation during the entire Meiyu stage.The formation of process I and II could be attributed to meeting of northward and southward air flows， while the essence of process III was typhoon-induced convection.The spatiotemporal variation of large-scale circulation associated with different processes was investigated comprehensively， and the result indicated that compared to process I， precipitation intensity of process II was very limited because of the forceless cold air invasion， more powerful SCSSM， and feeble transport of southwest moisture.Furthermore， the impact of precursory oceanic， atmospheric， and land signals on Meiyu was analyzed， and found that the primary climate characteristics were positive Ni?o index and negative North Pacific sea surface temperature （SST） anomaly in winter and spring， warm SST over southwest Indian Ocean in winter， decreased sea ice for the Kara-Barents Sea in spring， and Southern Annular Mode and Arctic Oscillation in the positive and negative phases respectively during April-May.Based on the monitoring time-series of Zhejiang Meiyu， the stable or strengthened climatic factors from winter to spring were extracted with two indices of correlation coefficient and ratio of the same sign， and moreover， winter and spring Meiyu prediction model were constructed using three methods of multivariable linear regression， multi-factors composite assess， and combined diagnosis.Generally speaking， all the linear statistical models were able to predict Meiyu precipitation anomaly over Zhejiang province well， especially for the negative anomaly cases.