To assess the relations of vascular ground flora to soil organic carbon (SOC) contents, we carried out plant census and soil sampling in 3 forest stands, i.e. 1-year and 3-year eucalyptus planted forests and a secondary broadleaved forest in western Guangdong, China. Variability in species composition and diversity in the ground flora was tested using multi-response permutation procedures (MRPP), and potential species indicators were detected by indicator species analysis (ISA). Dominant species in the ground flora differentiated natural secondary forest from eucalyptus planted forests, and the SOC content of the 0~25 cm soil layer varied significantly across the 3 forest stands. Significant differences in SOC content were found among the 5 groupings of the ground flora, which was defined by two-way indicator species analysis (TWINSPAN). MRPP indicated that species composition and diversity of the ground flora varied significantly in response to a gradient of SOC content. ISA revealed that 19 species, including Smilax glabra (IV > 50), were indicative of soil environments with high SOC content (> 30 g·kg^-1), and only 1 species, i.e. Inula cappa, were indicative of low SOC content (< 20 g·kg^-1). This study suggested that indicator species of different SOC regimes can be detected using a simple community analysis technique, and such indicator plants can be used for soil nutrient and soil carbon pool dynamics monitoring and assessment.