This study presents results of comparisons between instrumental radiation data in Tibet and two satellite products: the Global Energy and Water cycle Experiment - Surface Radiation Budget (GEWEX-SRB) and International Satellite Cloud Climatology Project - Flux Data (ISCCP-FD). Severe and systematic errors were found in monthly-mean SRB shortwave radiation (SW) and FD longwave radiation (LW). Errors in diurnal variations are even larger than the monthly mean errors. Though the LW errors can be reduced nearly 10 W m-2 after an altitudinal correction, the remaining errors are still higher than that for other regions. The large errors in SRB SW were caused by neglecting elevation effects, but the errors in FD LW were mainly due to significant errors in input data. We suggest reanalyzing all ancillary data and reprocessing these satellite surface radiation data, at least for highland areas like Tibet.