Abstract: For more efficiently applying field operation and management of precision irrigation, an improved ant colony clustering algorithm was used to delineate irrigation management zones. Ant colony algorithm with the characteristics of discreteness and parallelism is applicable to data feature clustering. However when the data quantity is huge, ant colony clustering will take long time on data search and cause high computational complexity in the process of system circulation. Thus, for the purpose of decreasing computational complexity and accelerating clustering, initial clustering center was taken as the initial food source in the paper to guide ant colony to reduce the blindness of ant walking. Soil physical properties were taken as the data sources. After principal components analysis was used to eliminate correlations among initial indexes, improved ant colony clustering was performed to delineate site-specified irrigation management zones. According to the comparison of delineation management zones between improved ant colony clustering and K-means clustering, management zones delineated by the former showed the features that soil physical properties had stronger uniformity within the subzone and more significant difference between subzones. Delineation result based on the improved ant clustering indicated that the study area could be partitioned into two irrigation management zones. Soil field capacity, saturation moisture content and permanent wilting point in Zone I were greater than those in Zone II, which indicated that soil in Zone I had stronger drought resistance than that in Zone II under the same climate conditions. Delineation of irrigation management zones could provide references and data support for site-specified irrigation management.