The prediction of reference crop evapotranspiration (ET0) is great significant for crop water requirement calculation and field water management, which can provide an important scientific basis for agricultural water conservation and efficient use of water resources. Based on the day-by-day meteorological data of six meteorological stations in the Sanjiang Plain from 1961 to 2010, the Penman-Monteith formula was used to calculate ET0 and analyze the spatiotemporal characteristics of ET0 and related meteorological elements from 1961 to 2010; based on the NCEP reanalysis data and the output data of the daily series of the CanESM2 forecast factor of the atmospheric circulation model, the statistical downscaling model (SDSM) was used to predict ET0 under two emission scenarios, RCP4.5 and RCP8.5. The results showed that the ET0 from 1961 to 2010 showed an increasing trend, the multi-year annual mean temperature and ET0 trend were the same, while the annual mean wind speed, relative humidity and net radiation showed an overall decreasing trend, and the spatial distribution of multi-year annual mean ET0 showed a general trend that the central part was higher than the periphery, and the western part was higher than the eastern part; in terms of simulation accuracy test, the ET0 simulation values of “historical” simulation under CanESM2 model and the calculated values of the P-M formula corresponded to the Nash efficiency coefficient (NSE) range of 0.46~0.61 and the coefficient of determination R2 range of 0.53~0.61 for the regular + validation period (1961—2005), which implied that the SDSM simulation was effective. The trends of the monthly average daily values of ET0 in the three future time periods of 2011—2040, 2041—2070, and 2071—2100 under the two scenarios in the future 2011—2100 intra-annual ET0 changes were relatively consistent, all resembling a parabola with a downward opening, with May-July significantly higher than the level of the historical period (1961—2010), January-April and August slightly higher than the historical period, and September-December gradually converged with the historical period. The future change of ET0 between 2011—2100 would on an upward trend compared with the historical period, and the three time periods of 2011—2040, 2041—2070, and 2071—2100 under RCP4.5 scenario would be increased by 11.11%, 18.70%, and 20.24%, respectively, compared with the historical period, with the multi-year ET0 in the time period of 2011—2040 on a more obvious upward trend. The overall downward trend would slower in the 2041—2070 and 2071—2100 time periods; the three time periods under the RCP8.5 scenario would be increased by 13.01%, 24.05%, and 34.46%, respectively, compared with the historical period, and the multi-year ET0 would on the rise in all three time periods. The future increase of ET0 in the study area may lead to aggravation of water shortage problem, and the results of the study may provide scientific reference for optimal water resources management and irrigation system formulation in the study area.