Species are the fundamental units of biology, ecology and conservation, and progress in these fields is therefore hampered by widespread taxonomic bias and uncertainty. Numerous operational techniques based on molecular or phenotypic data have been designed to overcome this problem, yet existing procedures remain subjective or inconsistent, particularly when applying the biological species concept. We address this issue by developing quantitative methods for a classic technique in systematic zoology, namely the use of divergence between undisputed sympatric species as a yardstick for assessing the taxonomic status of allopatric forms. We calculated mean levels of differentiation in multiple phenotypic characters – including biometrics, plumage and voice – for 58 sympatric or parapatric species‐pairs from 29 avian families. We then used estimates of mean divergence to develop criteria for species delimitation based on data‐driven thresholds. Preliminary tests show that these criteria result in relatively few changes to avian taxonomy in Europe, yet are capable of extensive reassignment of species limits in poorly known tropical regions. While we recognize that species limits are in many cases inherently arbitrary, we argue that our system can be applied to the global avifauna to deliver taxonomic decisions with a high level of objectivity, consistency and transparency.