High levels of genetic variation enable species to adapt to changing environments and provide plant breeders with the raw materials necessary for artificial selection. In the present study, six AFLP primer pairs were used to assess the genetic diversity of Desmodium triflorum (L.) DC. from 12 populations in South China. A high percentage of polymorphic loci (P=76.16%) and high total gene diversity ($H_T=0.310$) were found, indicating that the genetic diversity of D. triflorum is high at the species level. Genetic diversity was also relatively high at the population level (P=55.85%, $H_e=0.230$). The coefficient of gene differentiation among populations ($G_{ST}$ was 0.255, indicating that while most genetic diversity resided within populations, there was also considerable differentiation among populations. AMOVA also indicated 24.29% of the total variation to be partitioned among populations (${Phi}_{ST}=0.243$). UPGMA clustering analysis based on genetic distances showed that the 12 populations could be separated into three subgroups: an eastern, a western, and a central-southern subgroup. However, a Mantel test revealed no significant correlation (r=0.286, p=0.983) between the geographical distances and genetic distances separating these populations; mountain barriers to gene flow and human disturbance may have confounded these correlations. The present study has provided some fundamental genetic data that will be of use in the exploitation of D. triflorum.