Using conventional meteorological data from 116 meteorological stations in Northeast China from 1951 to 2013, an information transfer model of daily reference evapotranspiration(ET0)between stations in Northeast China was constructed based on the information entropy theory, and the distribution of the information field was analyzed. The regional similarity structural characteristics of daily ET0 were analyzed using the cluster analysis method. Then, six representative stations were chosen based on the analysis. The rescaled range analysis method was used to analyze the time fractal characteristic of ET0 at the six stations. The results show that, in space, the information entropy of daily ET0 decreases with the increase of latitude in Northeast China, and the information transfer index of each station, with significant anisotropy, decreases with the increase of distance between the basic station and the auxiliary station. The information transfer at the site level presents a general pattern of distribution from south to north and from east to west. In time, the mean annual ET0 presents a generally decreasing trend in Northeast China, and the ET0 of the six representative stations, i. e. , the Harbin, Shenyang, Chifeng, Jagdaqi, Kiamusze, and Hailar stations, will tend to stabilize within a certain time period in the future and show strong persistence.