The ‘mass extinctions’ at the end of the Pleistocene were unique, both in the Pleistocene and earlier in the geological record, in that the species lost were nearly all large terrestrial mammals. Although a global phenomenon, late Pleistocene extinctions were most severe in North America, South America and Australia, and moderate in northern Eurasia (Europe plus Soviet Asia). In Africa, where nearly all of the late Pleistocene ‘megafauna’ survives to the present day, losses were slight.

Ruling out epidemic disease or cosmic catastrophe, the contending hypotheses to explain late Pleistocene extinctions are: (a) failure to adapt to climatic/environmental change; and (b) extermination by human hunters (‘prehistoric overkill’).

This review focuses on extinctions in northern Eurasia (mainly Europe) in comparison with North America. In addition to reviewing the faunal evidence, the highly relevant environmental and archaeological backgrounds are summarized. The latest survival dates of extinct species are estimated from stratigraphic occurrences of fossil remains, radiocarbon dates, or association with archaeological industries.

The Middle and Upper Pleistocene (ca. 700000–10000 BP) in northern Eurasia and North America was a time of constantly changing climate, ranging from phases of extensive glaciation in cold stages, to temperate periods (interglacials). In the Lateglacial (ca. 15000–10000 BP), during which most extinctions occurred, there was a major reorganization of vegetation, mainly involving the replacement of open vegetation by forests. These changes were more profound than earlier in the Last Cold Stage, but similar in nature to vegetational changes that took place at previous cold stage/ interglacial transitions.

The archaeological record shows that humans have been present in Europe since the early Middle Pleistocene. The arrival in Europe ca. 35000 BP of ‘anatomically modern humans’, with their technologically more advanced upper palaeolithic industries, was a ‘quantum leap’ in human history.

Extinctions occurred throughout the European Pleistocene, but until the late Pleistocene most losses were replaced by the evolution or immigration of new species, and most of those lost without replacement were small mammals. In marked contrast, extinctions without replacement in the late Pleistocene were almost entirely confined to the largest mammals (> 1000 kg) and some medium-large species (100–1000 kg). Late Pleistocene extinctions in northern Eurasia were not synchronized, but occurred in two broad phases: (a) ‘interglacial survivors’, e.g. Palaeoloxodon antiquus, which retreated to southern Europe prior to their disappearance before ca. 30000 BP, i.e. before the main glaciation; and (b) cold-adapted species (e.g. Mammuthusprimigenius; Megaloceros giganteus) that disappeared in the Lateglacial, at various times between ca. 14000 and 10000 BP. Even within a species, populations became extinct earlier in some areas than in others, e.g. the possible survival of M. primigenius in north-central Siberia ca. 2000 years later than in Europe.

In North America many more species were lost than in northern Eurasia, including many medium-large mammals in addition to the largest forms. At least for the commoner species, extinctions apparently all occurred within a much narrower time span, ca. 10500–11 500 BP, probably much less.

Any extinction hypothesis must explain why losses in North America were severe and sudden, whereas those in northern Eurasia were moderate and staggered. The close correlation of North American extinctions with the arrival of Clovis hunters south of the ice sheets ca. 11 500 BP is consistent with overkill. However, there is no such correlation for northern Eurasia, where most extinctions also occurred in the Lateglacial, more than 20000 years after the appearance of upper palaeolithic humans. Although Lateglacial climatic/environmental changes correlate with extinctions in North America and northern Eurasia, the climatic hypothesis neither explains why extinction patterns were so different in the two regions, nor why similar losses did not take place at previous cold stage/interglacial transitions.

From the evidence reviewed here, human predation at times of major climatic/ environmental change is suggested as the most probable cause of late Pleistocene extinctions. In northern Eurasia overkill became possible only when large-mammal distributions, and thus populations, were already severely reduced by such changes. Similar extinctions did not occur earlier in the Pleistocene because ‘anatomically modern humans’ with upper palaeolithic hunting technologies were not present. In North America the main reason that losses were severe and sudden is probably the close coincidence of Lateglacial climatic/environmental changes with the arrival of Clovis hunters.

The necessity for amassing a much greater body of accurate faunal, environmental and archaeological data relevant to this intriguing question is emphasized. In particular, many more high-quality radiocarbon dates are required to determine the late Pleistocene history of extinct taxa in considerably more detail.