Most of the unknown animal biodiversity on earth is in groups of invertebrates that are hyperdiverse and abundant, yet poorly known (“dark taxa”). The study of these organisms requires a multi-faceted approach and methodologies designed to tackle large numbers of species and specimens. The scuttle flies (Diptera: Phoridae) are a classic example of a dark taxon and the focus of this thesis. Paper I is a molecular phylogeny of the phorid genus Megaselia based on one nuclear (28S rDNA) and three mitochondrial (ND1, COI and 16S) markers from 145 species of Nordic Megaselia. Molecular data was analysed with Bayesian analysis, maximum likelihood, and parsimony methods. Based on these results, and supporting morphological data, we divide Megaselia into 22 informal species groups, 20 of which fall into a monophyletic “core Megaselia”. We discuss implications for the future circumscription of Megaselia and associated genera. Paper II presents a pipeline for rapid and cost-effective species discovery using the Oxford Nanopore mobile sequencing technology MinION. This paper reveals the presence of ca. 650 species of Phoridae from a single Malaise trap placed in Kibale National Park, Uganda. Based on our data, we estimate that the phorid fauna of the Afrotropical region could be as high as 100 000 species: this figure dwarfs previous diversity estimates. The implications for species discovery and description are discussed, and a new species (Megaselia sepsioides sp. nov.) is described. Paper III outlines a large-scale integrative approach to species discovery and delimitation in hyperdiverse groups, exemplified using a dataset of 18 000 phorid flies from Sweden. COI minibarcodes (313 bp) were obtained for all specimens and classified into putative species using different clustering methods (objective clustering, Poisson tree process, automatic barcode gap discovery and refined single linkage). No clustering method was accurate enough to use for species delimitation without confirmation from additional data. We found that the stability of a cluster to change across genetic-distance thresholds and the genetic variation within a cluster both accurately predict clusters where morphology is likely to be incongruent with barcode data. With molecular clustering integrated with morphological validation, we found that we could examine less than 5% of specimens and still delimit all species fully and accurately. Paper IV addresses questions about the scuttle fly fauna of Sweden with data from 32 000 scuttle flies from 37 sites and 4 time periods. We estimate that the total Swedish fauna contains 652-713 (based on Chao 1 or CNE estimates, respectively) species of scuttle flies, 1.5 times the 372 species currently documented from Sweden. Ordination techniques show that scuttle fly communities are organized in a gradient across Sweden, which is well correlated with plant hardiness zones defined by the Swedish Horticultural Society. Hierarchical modelling of species communities (HMSC) reveals that phorid community composition is largely determined by climatic and temporal variables, but much of the variance remains unexplained by the models we explored. Comparison of our phorid data with that of species more commonly utilised for biodiversity assessments revealed that phorids may allow more fine-scaled analysis as they may exist in smaller ranges, and that they additionally may give unique patterns of distribution that are unlike those seen in other taxa.