In the present study, daily downwelling shortwave (Q S ) and longwave radiation (Q L ) data from one satellite and two hybrid products have been evaluated using Global Tropical Moored Buoy Array during 2001-2009 in the tropical oceans. Daily satellite data are used from the Clouds and Earth's Radiant Energy System (CERES) program. Data are obtained using Moderate Resolution Imaging Spectroradiometer (MODIS) (CM) aboard the Terra and Aqua satellites. Coordinated Ocean Research Experiments (CORE-II) and Tropical Flux data (TropFlux) are the other two hybrid products used in this study.The analysis shows that majority of Q S observations as well as derived products lie in 200-300 Wm −2 range in all the three tropical oceans. Both Q S and Q L in all products overestimated the majority of the observations. Yet, they underestimated the lower (0-100 Wm −2 ) values in Q S and higher (300-440 Wm −2 ) values in Q L . Majority of the Q L observations lie within 390-420 Wm −2 range, and CM slightly overestimated this observed distribution in the Pacific and the Atlantic Oceans. But, majority of the observations in the Indian Ocean lie within 420-450 Wm −2 range. This implies that the tropical Indian Ocean receives 30 Wm −2 more energy as compared to the tropical Pacific and the Atlantic in the form of downwelling longwave radiation. Daily observed Q S shows dominant seasonal cycle over the central, the eastern Pacific and the eastern Atlantic. On the other hand, the western Pacific, the central Atlantic and the Indian Oceans show intraseasonal variations. All products show this variation with high root-mean-square error (RMSE) values (Q S and Q L ) over the Indian Ocean than in the Pacific and the Atlantic Oceans. Downwelling radiation from CORE-II shows highest RMSE (for both Q S and Q L ) with least correlation coefficient (CC), and TropFlux has lowest RMSE and highest CC among all products in all three tropical oceans. CM has intermediate values of standard deviation, CC and RMSE. These results are not seasonally dependent, since the seasonal statistics are consistent with seasonal changes. Assuming that the SST is only driven by the downwelling shortwave and longwave fluxes, the errors associated with monthly SST can be as large as 0.2-0.3 (0.1-0.2) °C associated with errors in Q S (Q L ). Both Q S and Q L in CORE-II have lower spatial variability as compared to other datasets. Q L in the tropical oceans shows seasonal spatial variability determined by intertropical convergence zone positions. This variability does not change significantly over the Pacific and the Atlantic Oceans. The summer and winter monsoon patterns in the Indian Ocean guide the Q L variability. Opposite to Q S , higher Q L values have lower variability.Thus, this study aims at finding better radiation dataset to use in the numerical models and deduce that satellite data could be an alternative to existing reanalysis products.