The climate and vegetation for southern Primorye during the last 21 kyr are studied for the first time in detail using quantitative techniques on high-temporal resolution pollen record. The data obtained showed a clear relationship with global climate events: Last Glacial Maximum, Heinrich 1, Bølling/Allerød, Younger Dryas and the warmest climate event around ca. 8.0 ka. A moderate monsoon-type climate persisted in Primorye during the last 21 kyr. The mean annual temperature did not fall below 0°C. The magnitude of the glacial–interglacial mean temperature difference in southern Primorye was ca. 7°C. Cold and cool temperate climate existed between ca. 21.0 and 11.6 ka and after 4.3 ka respectively. Climate variables of the Last Glacial Maximum were slightly below present-day values, and drier conditions prevailed compared to the late Holocene. Moderately cool conditions with excessive moisture existed during the cold events of the late Pleistocene, i.e., Younger Dryas and Heinrich 1, and in the late Holocene. Humidification did not occur due to heavy rainfall, but due to less evaporation because of low average annual temperatures during these cold events. During the last 21 kyr, there existed dry and wet seasons in Primorye, as at present, associated with monsoon circulation. The summer monsoon was more intense at ca. 16.6 ka, 14.4 ka, 11.6 ka and from ca. 11.0 to 4.3 ka.The southern Primorye during last 21 kyr had a diverse spatial vegetation pattern with conifer and broadleaved deciduous forests of warmer or colder temperate type, dry and humid grasslands, open woodlands and riparian vegetation. Both today and in the past, the heterogeneity of the relief and climate created a diverse vegetation cover. The warmest component, broadleaved summergreen warm temperate trees, had a high proportion from 10.4 to 4.6 ka. Late Pleistocene plant communities, especially during the Last Glacial Maximum, were characterized by a high diversity of dry herbs, which indicates a dry steppe. There are indications of forest-steppe during LGM based on PFT and IPR analyses.