Confidence in molecular dating analyses have grown with the increasing sophistication of the methods. Previously problematic case studies where the molecular dates were in disagreement with palaeontological estimates appear to have been resolved. But we cannot relax just yet. The growing sophistication of molecular dating methods relies on an increasingly large number of assumptions about evolutionary history and processes. Many assumptions are based on statistical tractability rather than knowledge of molecular evolution, yet changing the prior assumptions can dramatically change the date estimates. How can we tell if the answers we get are driven more by the assumptions we make than by the molecular data being analysed?