Distribution of solar radiation is vital to locate the most suitable regions for harvesting solar energy, but solar radiation is only observed at few stations due to high costs and difficult maintenance. From 2001 to 2005, a set of pyranometer instruments were set up in Gaize, on the Tibetan Plateau, to test the hypothesis of high solar-radiation levels in this region, and find a suitable method for estimating the radiation. Over the 5-year observation period, the average daily radiation was 21 MJ m(-2)day(-1) with maximum daily values of 27 MJ m(-2)day(-1) occurring in June and minimum values of 14 MJ m(-2)day(-1) in December, which is much higher than those measured in other regions at similar latitudes. The observational data were used to validate a set of radiation models: five sunshine based and three temperature based. The results showed that of the five sunshine-based models, a newly developed "comprehensive" model performed the best, but that the "vapor revised Angstrom model" was recommended to use for its simplicity and easy operation. The temperature-based models performed worse than the sunshine-based ones, where the Wu model is to be preferred if a temperature-based model is the only option. Moreover, it was shown that when estimating the solar radiation based on time-dependent coefficients, consideration of the seasonal variation of the coefficients has little predictive value and is thus unnecessary. Based on the results of this study, a strategy for the calculation of solar radiation on the Tibetan Plateau was made for potential users. (C) 2011 Elsevier Ltd. All rights reserved.