The application of Stokes' formula requires that the atmospheric effect on
the gravity anomaly is removed. We show that this direct effect reaches abo
ut -40 cm over the Himalayas and Antarctica. The restoring of the atmospher
ic masses yields the indirect atmospheric effect, reaching about -20 cm for
the same regions. Consequently, the total atmospheric effect on the geoid
is of the order of - 60 cm. However, for most areas close to sea level, the
correction is within a few centimetres. Furthermore, the total atmospheric
geoid effect is derived for the truncated as well as the modified Stokes f
ormula. It is emphasized that the traditional (IAG) approach to adding a di
rect atmospheric effect to gravity may lead to a serious geoid bias in the
truncated Stokes formula. However, as all the parameters of the bias are kn
own, it can easily be corrected. In contrast, we suggest that the total atm
ospheric effect on the geoid be determined separately. In this approach the
bias is avoided.