Abstract Geometrid moths, the second largest radiation of Lepidoptera, have been the target of extensive phylogenetic studies. Those studies have flagged several problems in tree topology that have remained unanswered. We address three of those: (i) deep nodes of Geometridae (subfamilies Sterrhinae + Larentiinae, or Sterrhinae alone as sister to all other subfamilies), (ii) the taxonomic status of subfamily Orthostixinae and (iii) the systematic position of the genus Eumelea (classified in Desmobathrinae: Eumeleini or incertae sedis earlier). We address these questions by using a phylogenomic approach, a novel method on these moths, with up to 1000 protein‐coding genes extracted from whole‐genome shotgun sequencing data. Our datasets include representatives from all geometrid subfamilies and we analyse the data by using three different tree search strategies: partitioned models, GHOST model and multispecies coalescent analysis. Despite the extensive data, we found incongruences in tree topologies. Eumelea did not associate with Desmobathrinae as suggested earlier, but instead, it was recovered in three different phylogenetic positions, either as sister to Oenochrominae, Geometrinae or as sister to Oenochrominae + Geometrinae. Orthostixinae, represented by its type species, falls within Desmobathrinae. We propose the following taxonomic changes: we raise Eumeleini to subfamily rank as Eumeleinae stat. nov. and we treat Orthostixinae as a junior synonym of Desmobathrinae syn. nov.