Many species have mitochondrial DNA lineages that are phylogenetically intermixed with other species, but studies have rarely tested the cause of such paraphyly. In this study, we tested two hypotheses that could explain mitochondrial paraphyly of Holarctic gadwalls (Anas strepera) with respect to Asian falcated ducks (A. falcata). First, hybridization could have resulted in falcated duck mitochondrial DNA (mtDNA) introgressing into the gadwall gene pool. Second, gadwalls and falcated ducks could have diverged so recently that mtDNA lineages have not sorted to reciprocal monophyly. We used coalescent analyses of three independent loci to distinguish between these two hypotheses. Two lines of evidence support introgression. First, analyses of the three loci combined show that some introgression is necessary to explain current genetic diversity in gadwalls. Second, we generated alternative predictions regarding time since divergence estimated from mtDNA: falcated ducks and gadwalls would have diverged between 65,000 and 700,000 years before present (ybp) under the introgression hypothesis and between 11,000 and 76,000 ybp under the incomplete lineage sorting hypothesis. The two independent nuclear introns indicated that these species diverged between 210,000 and 5,200,000 ybp, which did not overlap the predicted time for incomplete lineage sorting. These analyses also suggested that ancient introgression (∼14,000 ybp) has resulted in the widespread distribution and high frequency of falcated-like mtDNA (5.5% of haplotypes) in North America. This is the first study to use a rigorous quantitative framework to reject incomplete lineage sorting as the cause of mitochondrial paraphyly.