According to the vegetation investigation and pollen analysis of surface samples sampled along a precipitation gradient of the Northeast China Transect (NECT), several pollen taxa, including Pinus, Betula, Quercus, Tilia, Acer, Ulmus, Artemisia, Chenopodiaceae, Gramineae and Cyperaceae, were chosen to make the regression and correlation analyses. The results indicated that there exists a close relationship between vegetation and pollen taxa in surface samples. The regression parameters for ten taxa in the forests in the eastern part of NECT were different from those in the steppes in the western part. Pinus, Betula, Artemisia and Chenopodiaceae, which have large slope and y-intercept terms, were over-representative taxa. Acer, Gramineae and Cyperaceae, which have small slope and y-intercept terms, were under-representative taxa. Quercus, Tilia and Ulmus whose slope terms have negative correlation with y-intercept terms were equi-representative taxa. The pollen taxa with large slope or large y-intercept terms have small variability coefficients, implying that the slope and y-intercept terms for these pollen taxa are of high accuracy in the estimation of plant abundance from pollen frequencies.