The selection of a DNA barcode in plants has been impeded in part due to the relatively low rates of nucleotide substitution observed at the most accessible plastid markers. However, the absence of consensus also reflects a lack of standards for comparing potential barcode markers. While many publications have suggested a host of plant DNA barcodes, the studies cannot be readily compared with each other through any quantitative or statistical parameter, partly because they put forward no single compelling rationale relevant to the adoption of a DNA barcode in plants. Here, we argue that the efficacy of any particular plant DNA barcode selection should reflect the anticipated performance of the resulting barcode database in assignment of a query sequence to species. While legitimate scientific disagreement exists over the criteria relevant to “database performance”, the notion gives a unifying rationale for prioritizing selection criteria. Accordingly, we suggest a measure of barcode efficacy based on the rationale of database performance, “the probability of correct identification” (PCI). Moreover, the definition of PCI is left flexible enough to handle most of the scientific disagreement over how to best evaluate DNA barcodes. Finally, we consider how different types of barcodes might require different methods of analysis and database design and indicate how the analysis might affect the selection of the most broadly effective barcode for land plants.