Abstract: In order to comprehend the distribution patterns and recovery status of the vegetation at Sanxingdui City Wall, this study focused on the herbaceous plant community to investigate the impacts of five maintenance measures. A field survey was carried out on five typical herbaceous plant communities associated with five common maintenance measures at Sanxingdui City Wall: natural regeneration, planting, abandoned field, shrub removal, and pruning. The survey examined species diversity, biomass and their interrelationships. A total of 103 herbaceous plant species from 40 families and 81 genera were identified in the study area, with perennial herbaceous plants dominating each site type, and Asteraceae and Gramineae as the dominant families. The highest values for the Shannon-Wiener diversity index, Pielou evenness index, species richness index, and Simpson dominance index were observed in the shrub removal plot, while the lowest values were in the planting plot; no significant differences were found among the four diversity indices in the control plot, abandoned field, shrub removal plot and pruning plot (P>0.05). The distribution of biomass, both above-ground and below-ground, varied significantly among the experimental plots. Notably, the control plot exhibited the highest biomass levels, followed by the planting plot, while the abandoned field showed the lowest biomass accumulation. Delving deeper into root stratification, it was observed that root biomass decreased as the soil depth increasing. In the topsoil layer (0—20 cm), the planting plot boasted the highest root biomass, contrasting with the pruning plot that had the lowest. Moving down to the 20—40 cm layer, it was intriguing to find that the abandoned field displayed a notably higher root biomass compared to other plots (P<0.05). However, no significant disparities were detected in root biomass among different plot types in both the 10—20 cm and 40—60 cm soil layers. The aboveground biomass of herbaceous plants exhibited significantly negative correlation with the Shannon-Wiener diversity index, Simpson dominance index, and Pielou evenness index (P<0.01), while showing no significant correlation with the species richness index (P>0.05). Both belowground biomass and total biomass did not demonstrate significant correlations with any of the four diversity indices. In summary, the original habitat protection measures mainly focused on natural regeneration and were supplemented by targeted interventions such as shrub removal and pruning. At the same time, the proportion of aboveground and underground biomass was controlled to slow down the destructive effects of root splitting and biological weathering, promote species diversity and biomass stability in the range most conducive to site protection and display.