Oxygen isotope records and glaciomarine sediments indicate at least an intermittent presence of large continental ice sheets on Antarctica since the earliest Oligocene (circa 35 Ma). The growth and decay of ice sheets during the Oligocene to modern “ice house world” caused glacioeustatic sea level changes. The early Eocene was an ice-free “greenhouse world,” but it is not clear if ice sheets existed during the middle to late Eocene “doubt house world.” Benthic foraminiferal δ18O records place limits on the history of glaciation, suggesting the presence of ice sheets at least intermittently since the earliest Oligocene. The best indicator of ice growth is a coeval increase in global benthic and western equatorial planktonic δ18O records. Although planktonic isotope records from the western equatorial regions are limited, subtropical planktonic foraminifera may also record such ice volume changes. It is difficult to apply these established principles to the Cenozoic δ18O record because of the lack of adequate data and problems in stratigraphic correlations that obscure isotope events. We improved Oligocene to Miocene correlations of δ18O records and erected eight oxygen isotope zones (Oi1-Oi2, Mi1-Mi6). Benthic foraminiferal δ18O increases which are associated with the bases of Zones Oil (circa 35.8 Ma), Oi2 (circa 32.5 Ma), and Mil (circa 23.5 Ma) can be linked with δ18O increases in subtropical planktonic foraminifera and with intervals of glacial sedimentation on or near Antarctica. Our new correlations of middle Miocene benthic and western equatorial planktonic δ18O records show remarkable agreement in timing and amplitude. We interpret benthic-planktonic covariance to reflect substantial ice volume increases near the bases of Zones Mi2 (circa 16.1 Ma), Mi3 (circa 13.6 Ma), and possibly Mi5 (circa 11.3 Ma). Possible glacioeustatic lowerings are associated with the δ18O increases which culminated with the bases of Zone Mi4 (circa 12.6 Ma) and Mi6 (circa 9.6 Ma), although low-latitude planktonic δ18O records are required to test this. These inferred glacioeustatic lowerings can be linked to seismic and rock disconformities. For example, we link 12 Oligocene-early late Miocene inferred glacioeustatic lowerings with 12 of the sequence boundaries (= inferred eustatic lowerings) of Haq et al. (1987).