<p>Paleomagnetic information recorded in lacustrine and marine sediments has been widely used in reconstructing the geomagnetic field on multi-millennial timescales. Whether a sediment has undergone inclination flattening, arising from sedimentary compaction, has not yet been systematically considered. In this study, we evaluate 82 individual sediment records by first comparing the paleomagnetic directions of the records to the THG24 field model, which is based on lavas over the past 10 Ma, to assess whether they show a similar dispersion in directions. Most records do not show the same dispersion and thus have a source other than paleosecular variation. These records are then tested for inclination flattening using the Secular Variation Elongation/Inclination (SVEI) method; only one record indicates flattening. Applying the original E/I method suggests flattening at 27 sites, located at mid-latitudes. The inclination data, which is corrected for flattening, is then used in the GGF100k model to show how the correction affects the axial and non-axial components of the geomagnetic field. The octupole terms are most affected by the correction, particularly the terms <i>g</i><Stack> ₃ ⁰ </Stack>, <i>h</i><Stack> ₃ ² </Stack>, and <i>h</i><Stack> ₃ ¹ </Stack>. In summary, this study implies that, based on the recently developed SVEI method, no compaction-induced inclination shallowing behaviour occurred over the past 100 kyr. Notably, non-dipole features of the geomagnetic field are an important part of the Earth’s magnetic field during this time period, underscoring the necessity of employing models that consider non-dipole terms in correcting inclination flattening. Furthermore, the 100 kyr data set displays more dispersion than paleomagnetic data from lavas over the past 10 Ma.</p>