The Gravity Recovery and Climate Experiment (GRACE) satellite mission will resolve temporal variations in gravity orders of magnitude more accurately and with considerably higher resolution than any existing satellite. Effects of atmospheric mass over land will be removed prior to estimating the gravitational field, using surface pressure fields generated by global weather forecast centers. To recover the continental hydrological signal with an accuracy of 1 cm of equivalent water thickness down to scales of a few hundred kilometers, atmospheric pressure must be known to an accuracy of 1 mbar or better. We estimate errors in analyzed pressure fields and the impact of those errors on GRACE surface mass estimates by comparing analyzed fields with barometric surface pressure measurements in the United States and North Africa/Arabian peninsula. We consider (1) the error in 30-day averages of the pressure field, significant because the final GRACE product will average measurements collected over 30-day intervals, and (2) the short-period error in the pressure fields which would be aliased by GRACE orbital passes. Because the GRACE results will average surface mass over scales of several hundred kilometers, we assess the pressure field accuracy averaged over those same spatial scales. The atmospheric error over the 30-day averaging period, which will map directly into GRACE data, is generally < 0.5 mbar. Consequently, analyzed pressure fields will be adequate to remove the atmospheric contribution from GRACE hydrological estimates to subcentimeter levels. However, the short-period error in the pressure field, which would alias into GRACE data, could potentially contribute errors equivalent to 1 cm of water thickness. We also show that given sufficiently dense barometric coverage, an adequate surface pressure field can be constructed from surface pressure measurements alone. Copyright 2001 by the American Geophysical Union.