AbstractLarval feeding by the moth genus Ogmograptis (Bucculatricidae: Lepidoptera) creates one of the most iconic features of the Australian bush—the ‘scribbles’ found on smooth‐barked Eucalyptus. The taxonomic history of Ogmograptis has been challenging, with members of the genus being initially described in four different genera representing three different superfamilies. While prior phylogenetic analysis has placed Ogmograptis within the Bucculatricidae, these findings were not strongly supported and there was poor resolution of the early diverging, non‐Apoditrysia superfamilies that Ogmograptis has been assigned to by different authors. As a consequence, the unique larval biology of scribbly moths cannot yet be interpreted in an evolutionary context. Phylogenomic analysis of whole mitochondrial (mt) genome data for Ogmograptis, related non‐Apoditrysia and taxa representing the superfamily‐level diversity of the order strongly supports its placement within the Bucculatricidae, a monophyletic Gracillarioidea and a clade of Gracillarioidea + Yponomeutoidea that was sister to the Apoditrysia. The hypermetamorphic larval development in Ogmograptis can thus be interpreted as an elaboration of the ancestral pattern of the clade Gracillarioidea + Yponomeutoidea that has specialised for phellogen/callus feeding within the bark. The utility of mt genomes for deep‐level phylogenetic study of the Lepidoptera is reviewed against prior multi‐locus and nuclear phylogenomic datasets. Mt phylogenomic analyses are sensitive to analytical methods and the inclusion versus exclusion of high‐variability data partitions for deep‐level relationships, already shown to be uncertain by multi‐locus or nuclear phylogenomic analyses, in particular relationships between apoditrysian and obtectomeran superfamilies. While mt genomes are ideal for examining the relationships of rare, physically small or difficult to collect taxa such as Ogmograptis, due to the low technical hurdles to collecting whole genomes, continued attention to the analytical sensitivities of phylogenies that use this data source is needed to reliably advance our understanding of deep lepidopteran evolution.