Reliable quantitative reconstructions of past vegetation require robust calibration of modern pollen-vegetation relationships across contrasting environmental settings. Such calibration remains limited in monsoon-influenced regions of northern China, where vegetation patterns reflect strong interactions among East Asian monsoon dynamics, forest-steppe transitions, and long-term human land use. The REVEALS (Regional Estimates of VEgetation Abundance from Large Sites) model has been widely applied for pollen-based land-cover reconstruction, yet its performance has been evaluated predominantly in Europe and only rarely under open-land dominated landscapes. Here we test the performance of the REVEALS model using modern pollen assemblages from 109 lake and reservoir surface sediments spanning major vegetation gradients across northern China. REVEALS-derived vegetation estimates were compared with contemporary land-cover maps within 51 sub-regions defined by a 50-km radius, and evaluated against uncorrected pollen proportions. The results show that REVEALS reconstructions closely reproduce the regional cover of the main vegetation categories, open land, broadleaved forest, and coniferous forest, and consistently provide more accurate estimates than pollen percentages at both individual-site and regional scales. Model performance is strongest when vegetation is aggregated into land cover groups rather than interpreted at the level of individual taxa. Our findings demonstrate that REVEALS provide robust quantitative vegetation estimates in monsoon-steppe ecosystems despite strong open-land dominance and heterogeneous topography. This study extends empirical validation of the REVEALS framework beyond Europe and supports its broader application for reconstructing past land cover and ecosystem dynamics in northern China and comparable semi-arid regions.