We investigated the effect of omitting the air cavities of the skull due to using a simplified model of the skull instead of a more realistic one (reference) in the dipole estimation. Our results showed that more than 95% of the localization errors made by the omission of the air cavities were below 3 mm.