Proposals for genetic thresholds for species delimitation assume that simple genetic datasets (e.g. mitochondrial sequence data) are correlated with speciation; i.e. that such datasets accurately reflect organismal lineage divergence. We used taxonomically stratified phenotypic levels of differentiation (populations, subspecies, and species) among nine avian lineages using paired, trans-Beringian samples from three lineages each in three orders (Anseriformes, Charadriiformes, and Passeriformes) to test this assumption. Using mitochondrial DNA (mtDNA) sequence data and nuclear genomic data (AFLPs), we found a lack of concordance between these two genomes in their respective estimates of divergence and little or no relationship between phenotype (taxonomic relatedness) and genetic differentiation between taxon pairs. There are several possible reasons for the discord observed (e.g. selection on one of the genomes or perhaps lineage sorting), but the implications are that genetic estimates of lineage divergence may not be correlated with estimates from other parts of the genome, are not well correlated with the speciation process, and are thus not reliable indicators of species limits.