Standard models of molecular evolution cannot estimate absolute speciation times alone, and require external calibrations to do so. Because fossil calibration methods rely on the unreliable fossil record, most nodes in the tree of life are dated with poor accuracy. However, many major paleogeographical events are dated, and since biogeographic processes depend on paleogeographical conditions, biogeographic dating may be used as an alternative or complementary method to fossil dating. I demonstrate how a time-stratified biogeographic stochastic process may be used to estimate absolute divergence times by conditioning on dated paleogeographical events. Informed by the current paleogeographical literature, I construct an empirical dispersal graph using 25 areas and 26 epochs for the past 540 Ma of Earth’s history. Simulations indicate biogeographic dating performs well so long as paleogeography imposes constraint on biogeographic character evolution. To gauge whether biogeographic dating may have any practical use, I analyze the well-studied turtle clade (Testudines) then assess how well biogeographic dating fares compared to heavily fossil-calibrated dating results as reported in the literature. Fossil-free biogeographic dating estimated the age of the most recent common ancestor of extant turtles to be approximately 201 Ma, which is consistent with fossil-based estimates. Accuracy improves further when including a root node fossil calibration. The described model, paleogeographical dispersal graph, and analysis scripts are available for use with RevBayes.