An astronomically calibrated timescale has recently been established <Hilgen, 1991a, b> for the Pliocene and earliest Pleistocene based on the correlation of dominantly precession controlled sedimentary cycles (sapropels and carbonate cycles) in Mediterranean marine sequences to the precession time series of the astronomical solution of Berger and Loutre <1991> (hereinafter referred to as Ber90). Here we evaluate the accuracy of this timescale by (1) comparing the sedimentary cycle patterns with 65 ¿N summer insolation time series of different astronomical solutions and (2) a cross-spectral comparison between the obliquity-related components in the 65¿ N summer insolation curves and high-resolution paleoclimatic records derived from the same sections used to construct the timescale. Our results show that the carbonate cycles older than 3.5 m.y. should be calibrated to one precession cycle older than previously proposed. Application of the astronomical solution of Laskar <1990> (hereinafter referred to as La90) with present-day values for the dynamical ellipticity of the Earth and tidal dissipation by the Sun and Moon results in the best fit with the geological record, indicating that this solution is the most accurate from a geological point of view. Application of Ber90, or La90 solutions with dynamical ellipticity values smaller or larger than the present-day value, results in a less obvious fit with the geological record. This implies that the change in the planetary shape of the Earth associated with ice loading and unloading near the poles during the last 5.3 million years was too small to drive the precession into resonance with the perturbation term, s6-g6+g5, of Jupiter and Saturn. Our new timescale results in a slight but significant modification of all ages of the sedimentary cycles, bioevents, reversal boundaries, chronostratigraphic boundaries, and glacial cycles. Moreover, a comparison of this timescale with the astronomical timescales of ODP site 846 <Shackleton et al., 1995a, b> and ODP site 659 <Tiedemann et al., 1994> indicates that all obliquity-related glacial cycles prior to ~4.7 Ma in ODP sites 659 and 846 should be correlated with one obliquity cycle older than previously proposed. ¿ American Geophysical Union 1996