A review of geochronological data underlying the geological timescale for the Triassic yields a significantly different timescale calibration than that published in the most recent compilation (Geologic TimeScale 2004). This is partly due to the availability of new radio–isotopic data, but mostly because strict selection criteria are applied and complications arising from biases (both systematic and random) are accounted for in this contribution. The ages for the base and the top of the Triassic are constrained by U–Pb ages to 252.3 and 201.5 Ma, respectively. These dates also constrain the ages of major extinction events at the Permian–Triassic and Triassic–Jurassic boundaries, and are statistically indistinguishable from ages obtained for the Siberian Traps and volcanic products from the Central Atlantic Magmatic Province, respectively, suggesting a causal link. Ages for these continental volcanics, however, are mostly from the K–Ar (40Ar/39Ar) system, which requires accounting and correcting for a systematic bias of c. 1% between U–Pb and 40Ar/39Ar isotopic ages (the 40Ar/39Ar ages being younger).Robust age constraints also exist for the Induan–Olenekian boundary (251.2 Ma) and the Early–Middle Triassic (Olenekian–Anisian) boundary (247.2 Ma), resulting in a surprisingly short duration of the Early Triassic, which has implications for the timing of biotic recovery and major changes in ocean chemistry during this time. Furthermore, the Anisian–Ladinian boundary is constrained to 242.0 Ma by new U–Pb and 40Ar/39Ar ages. Radio–isotopic ages for the Late Triassic are scarce, and the only reliable and biostratigraphically-controlled age is from an upper Carnian tuff dated to 230.9 Ma, yielding a duration of more than 35 Ma for the Late Triassic. All of these ages are from U–Pb analyses applied to zircons with uncertainties at the permil level or better. The resulting compilation can only serve as a guideline and must be considered a snapshot, resolving some of the issues mainly associated with inaccurate and misinterpreted data in previous publications. However, further advances will require revision of some of the data presented here.
