AbstractPlastid genome and nuclear ribosomal DNA (nrDNA) arrays, proposed recently as “super‐barcodes,” might provide additional discriminatory power and overcome the limitations of traditional barcoding loci, yet super‐barcodes need to be tested for their effectiveness in more plant groups. Morphological homoplasy among Schima species makes the genus a model for testing the efficacy of super‐barcodes. In this study, we generated multiple data sets comprising standard DNA barcodes (matK, rbcL, trnH‐psbA, nrITS) and super‐barcodes (plastid genome, nrDNA arrays) across 58 individuals from 12 out of 13 species of Schima from China. No samples were correctly assigned to species using standard DNA barcodes and nrDNA arrays, while only 27.27% of species with multiple accessions were distinguished using the plastid genome and its partitioned data sets—the lowest estimated rate of super‐barcode success in the literature so far. For Schima and other taxa with similarly recently divergence and low levels of genetic variation, incomplete lineage sorting, hybridization or taxonomic oversplitting are all possible causes of the failure. Taken together, our study suggests that by no means are super‐barcodes immune to the challenges imposed by evolutionary complexity. We therefore call for developing multilocus nuclear markers for species discrimination in plant groups.