Abstract: [Yield estimation research has great significance in making food policy, economic planning and the scientific macro- control of food. Aiming at those problems that most previous linear remote sensing yield estimation models were mainly based on the research of NDVI and LAI, and the multicollinearity among those remote sensing parameters, this paper chose six cities: Shijiazhuang, Baoding, Handan, Xingtai, Hengshui and Cangzhou, which lie in southern of Hebei Province, as the study area. The accumulated values of the optimal temporal NDVI and LAI in the planting region and the different months’surface temperature during 2000 to 2008 were selected as yield estimation indexes. The paper put forward a modified formula to correct the NDVI and LAI. Then, the principal component analysis was carried out on all the remote sensing parameters, and four yield estimation models were established by the corresponding principal components and the actual yield data. The data of 2009 was used to verify the accuracy of the models. The results showed that R2 of the four models were between 0.714 and 0.818, and the precisions of all were above 93.0%. Among them, the model integrating all remote sensing parameters achieved the best imitative effect. And R2 was 0.818, the yield estimation accuracy reached 95%; the precision of the model introducing the component of temperature was higher than that only using NDVI and LAI as the parameters; and the deeper analysis of the influence of various parameters on the yield estimation showed that the day-and-night surface temperature difference in the mid-April and the mid-May would have a great influence on the yield of winter wheat.