Lava flows and other igneous rocks become magnetized in the direction of the local geomagnetic field when they cool down after their formation. Similarly, sediments acquire a weak magnetic polarization on deposition. The mean direction of magnetization of series of recent lava flows and sediments has been determined; it is found that these mean directions agree closely with the theoretical dipole field. This field is the field due to a geocentric axial magnetic dipole. The conclusion is therefore drawn that the mean position of the magnetic poles (taken over a period of several thousand years) coincides with the geographic poles. Assuming that the same is true for the geological past, the position of the geographic poles can be defined within fairly narrow limits by using the mean direction of magnetization of older rocks. Measurements on igneous rocks, of Tertiary and Quaternary age and from Europe only, are available. It is concluded that the large amount of polar wandering suggested by Kreichgauer, Köppen and Wegener, and Milankovitch cannot be reconciled with the new data. If polar wandering has taken place at all, it has not exceeded 5°-10° since Eocene times.