We present the first gridded and temporally continuous quantitative pollen-based plant-cover reconstruction for temperate and northern subtropical China over the Holocene (11.7 ka to present) obtained by applying the Regional Estimates of VEgetation Abundance from Large Sites (REVEALS) model. The objective is to provide a dataset of pollen-based land cover for the last ca. 12 millennia that is suitable for palaeoclimate modelling and for the evaluation of simulated past vegetation cover from dynamic vegetation models and anthropogenic land-cover change (ALCC) scenarios. The REVEALS reconstruction was achieved using 94 selected pollen records from lakes and bogs at a 1× 1 spatial scale and a temporal resolution of 500 years between 11.7 and 0.7 ka and in three recent time windows (0.7-0.35 ka, 0.35-0.1 ka, and 0.1 ka to present). The dataset includes REVEALS estimates of cover and their standard errors (SEs) for 27 plant taxa in 75 1 × 1 grid cells distributed within the study region. The 27 plant taxa were also grouped into 6 plant functional types and 3 land-cover types (coniferous trees CT, broadleaved trees BT, and C3 herbs/open land (C3H/OL)), and their REVEALS estimates of cover and related SEs were calculated. We describe the protocol used for the selection of pollen records and the REVEALS application (with parameter settings) and explain the major rationales behind the protocol. As an illustration, we present, for eight selected time windows, gridded maps of the pollen-based REVEALS estimates of cover for the three land-cover types (CT, BT, and C3H/OL). We then discuss the reliability and limitations of the Chinese dataset of Holocene gridded REVEALS plant cover, and its current and potential uses. The dataset is available at the National Tibetan Plateau Data Center (TPDC; Li, 2022; 10.11888/Paleoenv.tpdc.272292).