Abstract: The study aims to inform agriculture administrators of winter wheat yield in advance, thus formulate food production form and relevant policy. Experimental fields of winter wheat yield data in May before harvest, and the synthetic aperture radar(SAR) double polarized—RDARSAT-2 that were photographed in April and May in Shouxian and Huaiyuan County in Huaihe River Region were adopted for winter wheat yield estimating. The liner yield estimating models were established by adopting backscattering coefficient and experimental fields yield. Then precision of yield estimating models of 2013 and 2014 were compared. Yield estimating models established by co-polarized (HH) and cross-polarized (HV) were adopted for winter yield estimating in Shouxian County, their precision were 68.37% and 74.01%, respectively, and their precision were 63.10% and 69.10% in Huaiyuan County, respectively. Yield difference of winter wheat in the lodging region was analyzed in detail. The precision of yield estimating model based on cross-polarized (HV) was higher than that based on co-polarized (HH). Finally, the four polarized SAR image would be chosen for yield estimating according to different winter wheat growth patterns. The results of this model analysis could form the research base and gather experience for parameter correction and popularizing of winter wheat yield models.