Population statistics of planktonic foraminiferal samples through DSDP Site 284 (Late Miocene to Recent) indicate surface-water temperature variations. According to biostratigraphic correlations, the temperature variations were less intense than, but were synchronous with inferred temperature variations on the eastern side of the North Island of New Zealand. The biostratigraphy permitted dating by reference to the paleomagnetic sequences determined in New Zealand marine strata. Water temperatures were relatively warm until t = 4.7 m.y. and were significantly cooler between t = 4.7 to t = 4.3 m.y. (equivalent to Kapitean Stage of New Zealand). A brief minor warming is associated with the New Zealand Miocene-Pliocene boundary and is followed by a cool interval to t = 4.1 m.y., a prolonged warm interval to t 2.6 m.y. (Opoitian Stage), and a significantly cooler episode to t = 2.15 m.y. Following this, temperatures fluctuate to an Early to Mid Pleistocene disconformity. The faunally determined paleotemperature changes coincide with changes in oxygen isotope composition of the water considered by Shackleton and Kennett (this volume) to indicate significant changes in ice volumes on Antarctica in the Late Miocene and within the Northern Hemisphere in the Late Pliocene and Pleistocene. The validity of planktonic foraminiferal paleotemperatures, placed in doubt by contradictory molluscan data in California and New Zealand, is justified by the comparative simplicity of planktonic ecologies, consistent sympathetic variations of different parameters of planktonic foraminiferal populations, and their consistent relationship to oxygen isotope data and paleoglacial history. Coiling of Neogloboquadrina pachyderma did not adopt its present-day pattern until early in the Pliocene and cannot be used for paleotemperature analysis of the Late Miocene and Early Pliocene faunas.