Abstract. The Drosophila cardini group includes 15 species, which are subdivided into the D. cardini and D. dunni subgroups. Although many phylogenetic hypotheses have been proposed for this group during the last five decades (based on patterns of reproductive isolation, morphology of male genitalia, chromosomal inversions, isozyme variation, or molecular sequence data), these are mostly discordant with each other. We aimed to clarify some of the evolutionary patterns related to the origin of this incongruence, while also attempting to provide a better‐supported phylogenetic hypothesis for the D. cardini subgroup. For this purpose, sequences from three mitochondrial and three nuclear loci were gathered for at least eight species, and both individual gene trees and joint species tree estimates were evaluated. Although there was concordance among gene trees within each of the nuclear and mitochondrial sets, considerable incongruence was revealed in the comparisons between these two data sets. The branching position of D. neocardini was the main source of incongruence, and species trees reconstructed using different approaches with and without this species were particularly incongruent. In addition to providing a better approximation of the evolutionary history of the D. cardini group, this study suggests that incomplete lineage sorting or introgression may be biasing previous species tree estimates. More generally, the results also suggest that the use of supermatrix methods can lead to an overestimation of support for the inferred relationships, and highlight the potential effects of different taxon sampling strategies in phylogenetic reconstruction.