Water holding capacity of forest litter is an important topic for forest eco-hydrology. Yet, most studies on forest litter water holding capacity are performed at plot or stand scales. There remains a lack of effective methods to extrapolate the forest water holding capacity from the plot or stand scale to larger spatial scale, e.g., watershed scale. By use of field sampling and experimental data on the water holding capacities of different types of forest stands and tree species composition along with remote sensing data e.g., vegetation index in the Manzizhuang watershed in Baotianman, this study established the spatial upscaling models based on forest vegetation structures (e.g., vegetation index and the proportions of different tree species) and environmental factors, and then applied the models to evaluate water holding capacity of forest litter and its spatial pattern at a watershed scale. We found that (1) The maximum water holding capacity of forest litter is significantly related to slope, tree height and the proportion of Quercus while the effective storage capacity is closely related to the average annual aboveground biomass and the proportion of Quercus. (2) The proportion of Quercus is a key factor for the water holding capacity of forest litter. The maximum water holding capacity and the effective storage capacity significantly increased by 59.2 and 55.6% (P<0.05), respectively, when the proportion of Quercus increased from 0 to 100%. Introducing the proportion of Quercus greatly promote the performance of the upscaling models of maximum water holding capacity and effective storage capacity forest litter with higher R² values of 0.918 and 0.875, respectively, and lower values of stand deviation (SD), root mean square error (RMSE) and mean absolute error (MAE). (3) According to the upscaling results, the average values of maximum water holding capacity and effective storage capacity of forest litter were 17.69 and 17.13 t/hm², respectively in Manzizhuang watershed. The areas with higher litter maximum water holding capacity were in the centre of the watershed, while greater litter effective storage capacity areas were located in the north and south of the watershed. (5) In terms of terrain and vegetation conditions, areas with greater litter water holding capacity were mostly featured with lower elevations and annual average aboveground biomass, higher maximum NDVI and the proportion of Quercus and flatter slopes. This study can provide scientific supports for forestry sectors to develop adaptive management measures for managing forest-water nexus.