Abstract: Abstract: In Tibet, radiation stations are scare. The radiation is important for accuracy estimation of reference evapotranspiration. In this study, the parameters of typical ?ngstr?m-Prescott (AP) equation were estimated and FAO recommended values were evaluated for its applicability in Tibet agricultural region. The data of daily solar radiation and sunshine hours from 1961 to 2016 were from all the 4 radiation stations (Lhasa, Shiquanhe, Nagchu and Chamdo stations) in Tibet. The AP parameters were fitted by nonlinear least square method. The data from the years of 1961 to 2004 were for calibration and those from the year of 2005 to 2016 were for validation. Standard deviation (SD), mean absolute error (MAE) and mean absolute percentage error (MAPE) were used for model accuracy evaluation in order to find the main factor affecting parameter change. In addition, the relationships between parameters and longitude, latitude and altitude were analyzed. The stations were in 2 climatic regions of plateau temperate semi-arid zone and plateau cold arid zone. According to the Tibet Agricultural Climatic Zoning, 5 zones were classified in Tibet, which included plateau temperate semi-humid zone, plateau cold arid zone, plateau temperate semi-arid zone, plateau warm-cool semi-humid zone and plateau temperate arid zone. Representative stations were chosen for these zones. Nyingchi, Shigatse, Suo county, Shiquanhe and Nagchu were for plateau temperate semi-humid zone, plateau temperate semi-arid zone, plateau warm-cool semi-humid zone, plateau temperate arid zone and plateau cold arid zone, respectively. Based on the relationship between main factor and parameters, the parameters for the other 3 zones were estimated by using inverse distance weighted method. The deviations from FAO recommended values were used for evaluation of FAO recommendations. The results showed that the SD, MAE and MAPE were in the acceptable range and the validation results for SD and MAE were 1.75-2.80 MJ/(m2·d), 1.27-2.03 MJ/(m2·d), respectively. The MAPE was closer or slightly different from 0. It indicated that the AP model was reliable for estimation of radiation. The calibrated coefficients for these 4 stations ranged from 0.18 to 0.26 and from 0.55 to 0.64, respectively. The recommended values by FAO were not always suitable in Tibet with the deviation of -28%-28%. The coefficient of variation of coefficient recommended by past studies was 15%-20%. Compared to previous studies with shorter time period, our results were more reliable since the time duration (about 50 years) was longer. The parameters were closely correlated with latitude with the determination coefficient higher than 0.6 but they were less related with longitude and altitude with the determination coefficient less than 0.2. Based on latitude, the interpolation was conducted for Nyingchi, Shigatse and Suo county. Finally the AP coefficients for 5 different agricultural zones were obtained and they ranged from 0.18 to 0.25 and from 0.55 to 0.64, respectively. The deviation analysis showed that the FAO recommended values had a good applicability in the plateau temperate semi-humid zone, plateau cold arid zone and plateau temperate semi-arid zone. The results can provide valuable information for the calculation of solar radiation and reference evapotranpiration in the areas without measured radiation data in Tibet.