We have constructed a very simple model of a time-varying geocentric dipole based on the archeomagnetic records obtained at four widely separated sites on the globe for the past 2 ka. The predictions of the model in terms of directional variations have been tested against actual archeomagnetic data from 12 sites distributed over the globe. They were also compared with the Hongre et al. ( 1998) time-varying spherical harmonic model and with the CALS7K-2 model by Korte et al. ( 2005). We find that the misfits between the predictions of our simple dipole model and the data are equivalent to those of the spherical harmonic models for the European sites and not strikingly larger for the rest of the world. Many discrepancies can be accounted for by uncertainties inherent to the archeomagnetic records, which, along with the small number and poor geographical distribution of sites, leads us to conclude that the present state of the database does not allow the extraction of secular variations described by terms going beyond degree 2. It appears also that dipole tilt could be responsible for the main part of the secular variation associated with time constants exceeding 10(2) years. In a second step, we used the paleointensity records contained in the same database to construct the curve depicting the variations of the true dipole moment. The present decrease of the dipole did not begin prior to 1000 years ago, and the dipole was actually increasing from 0 until A. D. 500. The dipole moment of CALS7K is lower than the present estimate, probably due to large repartition of energy to higher harmonics to minimize the misfit between the inversion and the data. The tilt and strength of the dipole can predict the dipole field at any site and were used to derive the contribution of the nondipole field to values of paleointensity at Paris during the past 2 ka. The results show that the "archeomagnetic jerks'' are associated with various configurations depending on the phase relationship between the nondipolar and dipolar parts of the field.