In order to implement the high precision of predicting reference evapotranspiration (ET0) when the meteorological information was not complete in northwest of China, FAO Penman—Monteith (P—M) formula was selected as control, 15 alternative methods were compared in Northwest China, including P—M formula with one to three factors of global solar radiation, sunshine hours duration, relative humidity, wind speed were missing, Pristley—Taylor（P—T）method, Makkink method, Hargreaves—Samani （H—S）method and Irmak method. At last, parameters of P—T method, H—S method, Makkink method and Irmak method were recalibrated. The results showed that global solar radiation data missing had the minimum impact on estimating ET0 with average R2 of 0.983 for all stations and the root mean square error (RMSE) was less than 0.4mm/d. When one of sunshine hour duration, wind speed or relative humidity data was missing, applying the FAO recommended P—M alternative methods could control RMSE within 0.47mm/d, R2 stayed above 0.94. When wind speed and relative humidity data were missing, Makkink method was the best method, the RMSE was 0.68mm/d and R2 was 0.94. When there was only temperature data, the improved Irmak and improved H—S methods had the same precisions with RMSE of 0.63mm/d and 0.68mm/d, the R2 were 0.94 and 0.95, respectively. In addition, the parameters of P—T, Makkink, H—S and Irmak methods were recalibrated for the Kriging interpolation, to obtain the spatial distribution of parameters. The required values of parameters for these methods can be got through the figures.