Abstract. The Dolichopodidae is a species-rich dipteran group with almost 7000 described species. The monophyly of the subfamilies and their relationships remain largely unknown because the polarities of key morphological characters are unclear and molecular data are available only for 9 of the 19 proposed subfamilies. Here we test whether molecular data from two nuclear (18S, 28S) and four mitochondrial (12S, 16S, Cytb, COI) genes can resolve the higher-level relationships within the family. Our study is based on 76 Oriental species from 12 dolichopodid subfamilies and uses eight species of Empididae and Hybotidae as outgroups. Parsimony and likelihood analyses confirm the monophyly of the Dolichopodidae, as well as the monophyly of five of the ten subfamilies represented by more than two species [Sympycninae, Sciapodinae, Dolichopodinae, Hydrophorinae (excluding tribe Aphrosylini), Neurigoninae]. There is strong support for restoring the tribe Aphrosylini as a separate subfamily Aphrosylinae. The monophyly of Medeterinae, Peloropeodinae and Diaphorinae is dependent on which tree reconstruction technique is used, how indels are coded, and whether the fast-evolving sites are excluded. Overall, we find that our sample of Oriental species is largely compatible with the subfamily concepts that were developed for the northern temperate fauna. However, our data provide little support for relationships between the subfamilies. Branch lengths, saturation, and distance plots suggest that this is probably the result of the rapid origin of dolichopodid subfamilies over a relatively short time. We find that genera that are difficult to place into subfamilies based on morphological characters are generally also difficult to place using molecular data. We predict that a dense, balanced taxon sample and protein-encoding nuclear genes will be needed to resolve the higher-level relationships in the Dolichopodidae.