Abstract: The crop water production function is of importance to water resource planners. The crop water sensitivity coefficient in this study is derived from the seasonal empirical model presented by Doorenbos and Kassam (1979). The coefficient is also known as yield response factor (Ky), which is an important basis for implementing efficient irrigation and optimal water allocation. Significant disparities in Ky are well documented on spatial scales. Although many related results for Ky under the condition of specific water management have been reported in previous literature, most of the studies have been focused on the value of Ky at an individual site, and few on its spatial variation or spatial patterns. This paper begins to study based on above problem. Therefore, this research has important theoretical significance and practical value.After determining the Ky values at the municipal level in the Haihe Basin, spatial statistical methods and exploratory spatial data analysis (ESDA) were implemented to examine the spatial pattern of Ky for winter wheat. Moran's I coefficient was used to study the global spatial autocorrelation, while Moran scatterplots and local indicators of spatial association (LISA) maps were used to study the local spatial autocorrelation of Ky. In addition, Moran's I values under different spatial directions were calculated to analyze spatial autocorrelation features in various directions including east-west (E-W), northeast-southwest (NE-SW), south-north (S-N) and southeast-northwest (SE-NW).Results showed that the Ky of winter wheat indicated an increasing trend from the western and northern mountainous region to the eastern plain in the basin, with values in the range 0.749~1.668. The global Moran's I values for dry, average, and wet typical growing seasons were 0.6009, 0.6058, 0.6077, respectively, and all with statistically significant differences (p<0.0001). This presents strong evidence of generalized, spatial autocorrelation between Ky at the global scale. The results of local spatial autocorrelation analysis revealed that Ky had a high-high (H-H) cluster in the eastern plain area including Beijing, Tianjin, etc., versus a low-low (L-L) cluster in the northern and western regions such as Chengde, Qinhuangdao, and Datong. Low-high (L-H) and high-low (H-L) clusters appeared to be rare. In addition, the total area in H-H and L-L clusters accounted for 80% of Haihe Basin, half of which exhibited statistical significance (p<0.05). Moreover, the degree of spatial autocorrelation of Ky diminished with the increasing of distance at each and every direction was accordance in trend, and the autocorrelation coefficient approached zero at a distance of 240~280 km. The NE-SW direction played a dominant role in spatial autocorrelation.In summary, the eastern plain area represents a "highly sensitive core region", diverging and decreasing gradually towards the western and northern mountains, forming a "core-periphery" spatial pattern. The research results could present some references valuable for water-saving irrigation and water resources optimal allocation in the Haihe Basin and provide effective clues for further study in other areas.