In order to investigate how reliably the carbon 14 content of tree rings reflects that of atmospheric carbon dioxide, two types of determinations were carried out: (1) carbon 14 determinations in annual rings from the beginning of this century until 1974 and (2) carbon 14 determinations in synchronous wood from the North American bristlecone pine and from European oak trees, dendrochronologically dated to have grown in the third and fourth century B.C. The first series of measurements showed that bomb-produced radiocarbon was incorporated in wood at a time when it was converted from sapwood to heartwood, whenever radiocarbon from bomb testing was present in the atmosphere. The second series showed that wood more than 2000 years old and grown on two different continents at different altitudes had, within the limits of experimental error, the same radiocarbon content. This work and other experimental evidence, obtained in part by other laboratories, show that tree rings reflect the average radiocarbon content of global atmospheric carbon dioxide accurately within several per mil. In rare cases, deviations of up to 10 0/00 may be possible. This means that a typical single radiocarbon date for wood or charcoal possesses an intrinsic uncertainty (viz., an estimated 'one-sigma error' in addition to all the other errors) of the order of ¿50 years. This intrinsic uncertainty is independent of the absolute age of the sample. More accurate dates can, in principle, be obtained by the so-called method of 'wiggle matching.'