Turtles have served as a model system for molecular divergence dating studies using fossil calibrations. However, because some parts of the fossil record of turtles are very well known, divergence age estimates from molecular phylogenies often do not differ greatly from those observed directly from the fossil record alone. Also, the phylogenetic position and age of turtle fossil calibrations used in previous studies have not been adequately justified. We provide the first explicitly justified minimum and soft maximum age constraints on 22 clades of turtles following best practice protocols. Using these data we undertook a Bayesian relaxed molecular clock analysis establishing a timescale for the evolution of crown Testudines that we exploit in attempting to address evolutionary questions that cannot be resolved with fossils alone. Some of these questions, such as whether the turtle crown originated in the Triassic or Jurassic, cannot be resolved by our analysis. However, our results generate novel age-of-origination estimates for clades within crown Testudines. Finally, we compare our fossil calibrations and posterior age estimates to those from other studies, revealing substantial differences in results and interpretation.