Although high-sensitivity liquid scintillation (LS) spectroscopy is theoretically capable of producing finite radiocarbon ages in the 50,000- to 70,000-yr range, there is little evidence in the literature that meaningful dates in this time period have been obtained. The pressing need to undertake calibration beyond 26 kyr has resulted in the regular publication of 14C results in excess of 50 kyr, yet very little effort has been made to demonstrate their accuracy or precision. There is a paucity of systematic studies of the techniques required to produce reliable dates close to background and the methods needed to assess contamination from either in situ sources or laboratory handling and processing. We have studied the requirements for producing accurate and reliable dates beyond 50 kyr. Laboratory procedures include optimization of LS spectrometers to obtain low and stable non-14C background count rates, use of low-background counting vials, large benzene volumes, long counting times, and preconditioning of vacuum lines. We also discuss the need for multiple analyses of a suitable material containing no original 14C (background blank) and the application of an appropriate statistical model to compensate for variability in background contamination beyond counting statistics. Accurate and reproducible finite ages >60 kyr are indeed possible by high-sensitivity LS spectroscopy, but require corroborating background blank data to be defensible.