Quantifying plant litter cover is important to evaluate the effectiveness of vegetation to protect soil against soil erosion. Field campaign was carried out in the Loess Hilly-gully Region of Northern Shaanxi to acquire the spectral reflectance data of plant-soil mixed scenes for two broadleaf forests and rehabilitated grassland. Spectral behavior of plant-soil mixed scenes was analyzed and the effectiveness of hyperspectral indexes NDLI (normalized difference lignin index) and CAI (cellulose absorption index) for quantifying plant litter cover was evaluated based on lab and field experiments. The results showed spectra reflectance of plant-soil mixed scenes with different proportions of litter and soil varied greatly by the influence of color and water content. The NDLI and CAI values increased with the rise of proportion of litter within the mixed scene under dry and wet status, regression analysis between the two indexes and plant litter cover of mixed scenes showed that the obviously better result for CAI (R2=0.98, rehabilitated grassland and broadleaf forests under dry status). The discrimination between mixed scenes and soil using CAI significantly outperformed NDLI under dry status; the estimation of plant litter cover by CAI is more effective compared to NDLI. Lab analysis was validated by field measuring: R2=0.90 showed highest correlation between CAI and plant litter cover for rehabilitated grassland. The validity of estimation of plant litter cover by both two indexes reduced to some degree in the field. The research enhanced the scientific basis for remote sensing estimation of plant litter cover.