The quality of the National Center for Environmental Prediction (NCEP)/National Center for Atmospheric Research (NCAR) reanalysis surface forcing fields is evaluated for the Bering Sea and the NE Pacific using a series of buoy measurements from 1995 to 2000. Because the mooring data have not been assimilated into the reanalysis, they provide an independent measure of the accuracy of the reanalysis in these regions. Emphasis is placed on 10 m winds and shortwave radiation, as they are important parameters for forcing ocean models. Except close to the coast where the reanalysis does not adequately account for the blocking influence of the topography, wind directions were accurately estimated by the reanalysis. Comparisons with the reanalysis suggest a small bias with NCEP 10 m winds greater than observed by ~5%. In the Bering Sea, downward shortwave radiation in the reanalysis is overestimated by roughly 70--80 W m-2 averaged over the summer data. In the NE Pacific, the reanalysis overestimates downward shortwave radiation by approximately 20 W m-2 in both summer and winter. Inaccurate representation of clouds in the reanalysis model is the most likely reason for the overestimate. Analysis of weather patterns during periods of particularly large overestimation of shortwave radiation suggests that the reanalysis better reproduces cloudiness during synoptic-scale cyclonic disturbances and is less able to accurately reproduce low clouds and sea fog in fair weather conditions.