Recent controversy about the use of the vertebrate fossil record for external calibration of molecular clocks centers on two issues, the number of dates used for calibration and the reliability of the fossil calibration date. Viewing matters from a palaeontological perspective, we propose three qualitative, phylogenetic criteria that can be used within a comparative framework for the selection of well-constrained calibration dates from the vertebrate fossil record. On the basis of these criteria, we identify three highly suitable new fossil calibration dates for molecular clock estimates: the lungfish–tetrapod split (between 419 and 408 Ma), the bird–crocodile split (between 251 and 243 Ma), and the alligator–caiman split (between 71 and 66 Ma). Together with our previously suggested bird–lizard split (between 252 and 257 Ma), these four fossil dates span a range of very different ages. They are, in our opinion, more suitable for molecular clock calibration than the traditionally used mammal–bird split, which is less well constrained. We plea for closer interactions between paleontologists and molecular biologists in studying the timescale of vertebrate evolution. BioEssays 27:1069–1075, 2005. © 2005 Wiley Periodicals, Inc.