Abstract This study had two aims. First, we tested the monophyly of and relationships within the ‘bombycoid complex’, an assembly of approximately 5300 species postulated by Minet to represent 12 families in three superfamilies, by sequencing five protein-coding nuclear gene regions (CAD, DDC, enolase, period, wingless; approximately 6750 bp total) in 66 representatives of most of the subfamilies and tribes. Second, we sought initial evidence on the utility of these genes for estimating relationships among Macrolepidoptera more broadly (11 superfamilies total), by adding representatives of eight families from four other superfamilies, and by assessing the phylogenetic information content of the individual genes and partitions thereof. Analysis of the combined data by likelihood and parsimony upholds monophyly for the bombycoid complex and for Bombycoidea sensu stricto (includes Anthelidae, see below), but with weak bootstrap support. Minet’s assignment of Phiditiinae to Bombycoidea rather than to Noctuoidea is strongly upheld, but Anthelidae, placed in Lasiocampoidea by Minet, group securely within Bombycoidea sensu stricto. Within the latter, the basal split segregates a strongly supported ‘BALE’ group [Apatelodinae + (Eupterotidae + (Brahmaeidae + Lemoniidae))]. The remaining families form a consistently but weakly supported clade, within which the basal split segregates the very strongly supported ‘CAPOPEM’ group [Carthaeidae, Anthelidae, Phiditiinae, (Prismostictini + (Endromidae + (Oberthueriini + Mirinidae)))]. The remaining bombycoids are grouped, very weakly, as Sphingidae + (Bombycinae + Saturniidae). All multiply-sampled families are strongly recovered, in both outgroups and ingroups, except that Bombycidae sensu Minet are rendered decisively polyphyletic. All genes make important contributions to the combined data results, and there is little strong conflict among genes or between synonymous and nonsynonymous change, although two instances of inter-gene conflict were notable, one in Lasiocampidae and one in Mimallonidae. Overall, about 75% of nodes are strongly supported (i.e. bootstrap value ≥80%). Superfamilies are recovered, but not always strongly, whereas relationships among superfamilies are recovered only weakly and inconsistently; even within the reasonably well-sampled Bombycoidea sensu stricto, a (to us) surprising number of interfamily relationships remain uncertain. Thus, it seems clear that substantially more genes, plus additional taxon sampling in most superfamilies, will be required to resolve macrolepidopteran phylogeny.