Samples of both archaeological and non-archaeological origins have been analyzed for ¹⁴C and the results used to assess the long-term dating reproducibility of this radiocarbon laboratory. A series of wood replicate analyses provides an estimate for this long-term precision of about 100 years, a value slightly in excess of the quoted counting error, typically 60-80 years. By a similar statistical treatment of the ¹⁴C data obtained from archaeological replicate and duplicate analyses performed over a three-year period, the resultant long-term reproducibility is shown to be in the range 75-120 years. Given the similarity in values obtained for each series, there is no additional variability apparent in the dating of archaeological samples, certainly at this level of precision. In contrast, short-term precision, over a few months, has been shown to be as low as 45 years. While this latter value underlines the benefit to ¹⁴C users of series counting of related samples using liquid scintillation counting, for the comparison of radiocarbon dates over several years, long-term estimates of dating precision are recommended.