After the mid-Holocene, human activities gradually began having a notable impact on land cover and plant compositional changes. Evaluating the extent and spatiotemporal variations of the human versus climate impacts on regional ecosystems is becoming an area of focus of current global change research. The present study uses 478 AMS 14C dating records collected from 5473 archaeological sites to help evaluate variations in prehistoric population size in Northeast China, which suggest changes in the nature of human activities there since the mid-Holocene. Results indicate that prehistoric human impacts remained at a relatively low level during the ca. 7–4 ka interval except for two minor fluctuations. Human impacts on ecosystems in the study area gradually intensified after 4 ka when societies entered the Bronze Age. In addition, we used a novel methodological approach on three pollen datasets for reconstructing the land cover and plant compositional changes of the three different studied landscapes (steppe, forest, and steppe-forest ecotone) in Northeast China. Results show that total land cover changes in forests were relatively low (i.e. stable) over the studied time period owing to their comparatively higher plant diversity whereas significant fluctuations occurred in the steppe and the steppe-forest ecotone. By comparing these results with regional climate records, climate change was found to dominate plant changes during two periods--before ca. 6 ka and after ca. 0.8 ka. In addition, during ca. 6–4 ka BP, even though climate still played the most significant role in vegetation changes, anthropogenic impacts on plant changes were revealed for the steppe-forest ecotone. During 4–2.3 ka BP, the anthropogenic impacts on plants gradually increased and became the dominant driving force, especially for the forest and steppe-forest ecotone during ca. 2.3–0.8 ka BP. These varied impacts of human activity and climate change on vegetation among the study areas can be related to human migration trends and human subsistence patterns.