The Gelechioidea (> 18 000 species), one of the largest superfamilies of Lepidoptera, are a major element of terrestrial ecosystems and include important pests and biological model species. Despite much recent progress, our understanding of the classification, phylogeny and evolution of Gelechioidea remains limited. Building on recent molecular studies of this superfamily and a recently revised family/subfamily classification, we provide an independent estimate of among-family relationships, with little overlap in gene sample. We analysed up to five nuclear genes, totalling 6633 bp, for each of 77 gelechioids, plus up to 14 additional genes, for a total of 14 826 bp, in 45 of those taxa and all 19 outgroup taxa. Our maximum-likelihood (ML) analyses, like those of previous authors, strongly support monophyly for most multiply-sampled families and subfamilies, but very weakly support most relationships above the family level. Our tree looks superficially divergent from that of the most recent molecular study of gelechioids, but when the previous tree is re-rooted to accord maximally with ours, the two phylogenies agree entirely on the deepest-level divergences in Gelechioidea, and strongly though incompletely on among-family relationships within the major groups. This concordance between independent studies is evidence that the groupings (or at least the unrooted branching order) are probably accurate, despite the low bootstrap values. After re-rooting, both trees divide the families into three monophyletic groups: a
Gelechiid Assemblage,' consisting of Gelechiidae and Cosmopterigidae; a Scythridid Assemblage,’ consisting of Stathmopodidae, Scythrididae, Blastobasidae, Elachistidae, Momphidae, Coleophoridae and Batrachedridae; and a
Depressariid Assemblage,' consisting of Autostichidae, Xyloryctidae, Lecithoceridae, Oecophoridae, Depressariidae and Lypusidae. Within the largest family, Gelechiidae, our results strongly support the pairing of Anomologinae with Gelechiinae, in accordance with a recent study of this family. Relationships among the other subfamilies, however, conflict moderately to strongly between studies, leaving the intrafamily phylogeny unsettled. Within the Scythridid Assemblage,’ both trees support an
SSB clade' consisting of Blastobasidae + (Scythrididae + Stathmopodidae), strongly resolved only in our results. Coleophoridae + Batrachedridae is supported, albeit weakly, in both trees, and only Momphidae differ in position between studies. Within the Depressariid Assemblage,’ both trees support an
AXLO' clade consisting of Autostichidae, Xyloryctidae, Lecithoceridae and Oecophoridae. The monophyly of this clade and relationships therein are supported weakly in previous results but strongly in ours. The recently re-defined family Depressariidae is paraphyletic in our tree, but the evidence against depressariid monophyly is very weak. There is moderate support for a core group of Depressariidae consisting, among the seven subfamilies we sampled, of Depressariinae, Aeolanthinae and Hypertrophinae. We show that gelechioids have a higher total number and percentage of species that are saprophagous as larvae than any other apoditrysian superfamily, that saprophagy is concentrated primarily in the AXLO clade,’ and that the ancestral gelechioid condition was probably feeding on live plants. Among the living-plant feeders, concealed external feeding was probably the ancestral state. The multiple origins of internal feeding of various kinds, including leaf mining (otherwise almost unknown in Apoditrysia), are restricted mostly to the Scythridid and Gelechiid Assemblages. The traits that predispose or permit lineages to adopt these unusual life histories are worthy of study.