Molecular and paleontological data provide independent means of estimating when groups of organisms evolved in the geological past, but neither approach can be considered straightforward. The single most fundamental obstacle to developing an accurate estimate of times of origination from gene sequence data is variation in rates of molecular evolution, both through time and among lineages. Although various techniques have been proposed to circumvent this problem, none unambiguously allow the components of time and rate to be separated. Furthermore, problems of establishing accurate calibration points, correctly rooted phylogenies, and accurate estimates of branch length remain formidable. Conversely, paleontological dates fix only the latest possible time of divergence, and so probabilistic methods are required to set a lower boundary on origination dates. Realistic confidence intervals that take preservational biases into account are only just becoming available. Although molecular and paleontological approaches to dating often agree reasonably well, there are two notable areas of disagreement; when mammal and bird orders originated and when the major phyla originated. The discrepancy in dating bird/mammal ordinal origins probably reflects a global rock-record bias. Paleontological sampling in the Late Cretaceous is still too restricted geographically to draw any firm conclusions about the existence of a pre-Tertiary record for modern orders of bird or mammal from anywhere other than North America. Dating the time of origin of phyla is more complicated, and is confounded by both preservational biases and problems of molecular clock estimation.