Using relative paleointensity estimates derived from twelve globally distributed pelagic sediment cores, we assess whether they record a signal consistent with that expected from a dominant geocentric axial dipole. The cores span the Matuyama-Brunhes boundary and we normalize the observations by supposing that at the time the direction reverses the intensity low reflects only the non-axial-dipole contribution to the field. We further assume that this non-axial-dipole contribution to the field is invariant with geographic location. From absolute paleointensity compilations we estimate its size to be about 7.5 μT; this supplies the calibration for the axial dipole signal away from the extreme low in intensity. The data predict the dipole field variation with latitude with similar accuracy to that observed in absolute paleointensity records, and show similar behavior when transformed to virtual axial dipole moments.