Discordance between entities revealed by nuclear versus mitochondrial genes is a common phenomenon in evolutionary and taxonomic studies. However, little attention has been paid to analysis of how such discordant entities correspond to traditional species detected through investigation of their morphology, ecology, and distribution. Here, we used one mitochondrial (COI, DNA barcode fragment) and four nuclear (CAD, Ca-ATPase, arginine kinase, wg) genes to analyze the genetic structure of the taxonomically well-studied butterfly genus Brenthis (Lepidoptera, Nymphalidae). Analysis of COI revealed multiple diverged allopatric and sympatric mitochondrial lineages within the known Brenthis species hinting at possible presence of unrecognized cryptic species. However, these multiple-species hypotheses were not supported by further studies of nuclear genes and phenotypic traits. The discovered mitochondrial lineages did not correspond to the clusters revealed by nuclear genes. Simultaneously, we found a complete congruence between (a) traditional species boundaries, (b) clusters recognized by nuclear genes, and (c) clusters identified via cladistic analysis of phenotypic traits (genitalia and wing pattern characters, ecological preferences, and chromosome numbers). We conclude that in case of the genus Brenthis, nuclear genes rather than mtDNA barcodes reveal real species boundaries. Additionally, we suggest to support each DNA barcode-based taxonomic conclusion by analysis of phased alleles of nuclear genes, avoiding widely used practice of nuclear and mitochondrial genes concatenation without any examination of interaction of these different types of data.