Abstract In this study we present a recent compilation of 286 modern surface pollen spectra from the southern part of the Russian Far East (42–54°N, 131–141°E) and use it to test the biome reconstruction method. Seventy terrestrial pollen taxa were assigned to plant functional types and then classified to eight regional biomes. When applied to 286 surface pollen spectra, the method assigns about 70% (201 sites) of the samples to the cool mixed forest biome, 17% – to the taiga, 2% – to the cool conifer forest, 3% – to the temperate deciduous forest, and 7% – to the steppe. The steppe reconstruction is characteristic of the pollen spectra from the agricultural areas around Lake Khanka. A visual comparison shows good agreement between pollen-derived biomes and actual vegetation distribution in the region. However, pollen and botanical data, compared with the potential vegetation distribution simulated from the modern climate dataset using the BIOME1 model, demonstrate that spatial distribution of cool mixed forest is underrepresented in the model simulation. The model sets the mean temperature of the coldest month of −15 °C as the factor limiting distribution of the temperate deciduous broadleaf taxa, while vegetation and pollen data from the region demonstrate that this limit should be lowered to −26 °C. Application of the method to the Gur 3–99 pollen record (50°00 ′N, 137°03 ′E) demonstrates that tundra vegetation predominated around the site prior to 14 ka BP (1 ka = 1000 cal. years). However, the local presence of boreal trees and mixed forest-tundra vegetation is suggested by relatively high taiga scores. Soon after 14 ka BP the scores of taiga become slightly higher than tundra scores. During 11.4–10.5 ka BP a cool conifer forest is reconstructed. Establishment of the full interglacial conditions is marked by the onset of cool mixed forest by 10.5 ka BP. Between 10.3 and 2.5 ka BP the scores of temperate deciduous forest are close to those of cool mixed forest and become distinctly lower during the late Holocene.