An analysis of spatial and temporal variations of the polar radiation budget will undoubtedly require the use of multispectral satellite data. How well we can estimate the radiation balance depends on how well we can estimate the physical and microphysical properties of the surface and atmosphere that directly affect it, e.g., surface temperature and albedo, cloud droplet effective radius, cloud optical depth, cloud thickness, and cloud height. Here we examine our current ability to estimate the high-latitude surface radiation budget using visible and thermal satellite data. The method for estimating radiative fluxes incorporates estimates of surface and atmospheric parameters, so the accuracy with which these can be retrieved from satellite data is first assessed. The effects of errors in the estimates of these parameters on the surface net radiation during summer and winter are quantified, and the relative sensitivity of the net radiation budget to errors in individual parameters is assessed. The combined uncertainty is then determined and examined in light of validation data in the Arctic. The results show upper and lower bounds for the uncertainties between 7.9 and 41 W m-2 for instantaneous retrievals of net radiation. By far, the largest portion of the uncertainty in net radiation is associated with errors in the retrieval of surface temperature and surface albedo. Although improvements in retrievals are desirable, currently available methods can provide surface net radiation in the Arctic with uncertainties similar to those of surface-based climatologies.¿ 1997 American Geophysical Union