Abstract: Solar radiation is often needed to be estimated by ordinary data in the current applications now, in which the sunshine-based Ångström-prescott formula is widely used with highest accuracy. However, the key of its application is to obtain the appropriate local parameters. In this paper, six parameterization methods, one from FAO and the other five from Liu et al., were evaluated using data from three sites in Yunnan province. The results showed that two methods of Liu et al. performed best, and their estimation accuracy was similar to the calibrated ones. In Kunming and Lijiang, the mean bias error (MBE), the root mean square error (RMSE) and the relative root mean square error (PRMSE) was -0.28-0.50 MJ/(m2·d), 2.09-2.39 MJ/(m2·d) and 12.4%-15.8%, respectively. In contrast, the MBE, RMSE and PRMSE obtained by the FAO scheme at these two sites were -0.43-0.84 MJ/(m2·d), 2.61-5.37 MJ/(m2·d) and 17.2%-31.7%, which showed a much lower accuracy than two methods of Liu et al. Considering the fact that it is still a common practice to directly use the FAO method, this research contributed to understand the importance of parameter choice for the Ångström-prescott model and suggested the calibrated or parameterized parameter can be used as much as possible in practice.