This paper investigates the sampling errors in the spherical harmonic coefficients of meteorological fields derived from observations made by polar-orbiting satellites. The sampling errors are attributed not only to the spatial gaps of satellite orbits but also to the nonsimultaneousness of satellite sampling. As a result, the mean-square error is found to consist of aliased powers from spatial waves as well as temporal frequencies. In addition to analytical results, a simple energy balance model is employed in a numerical study that produces interesting numerical results. In particular, the root-mean-square error for estimating the daily global average surface temperature anomaly using 16 polar orbits is found to be 0.06 ¿C or less if the temporal scale of decorrelation is equal to 30 days or longer; it can be as high as 1.12 ¿C if the temporal scale is decreased to 1 day. ¿ 1998 American Geophysical Union