It is commonly accepted that some of the latest dates for Neanderthal fossils and Mousterian industries are found south of the Ebro valley in Iberia at ca. 36 ka calBP (calibrated radiocarbon date ranges). In contrast, to the north of the valley the Mousterian disappears shortly before the Proto-Aurignacian appears at ca. 42 ka calBP. The latter is most likely produced by anatomically modern humans. However, two-thirds of dates from the south are radiocarbon dates, a technique that is particularly sensitive to carbon contaminants of a younger age that can be difficult to remove using routine pretreatment protocols. We have attempted to test the reliability of chronologies of 11 southern Iberian Middle and early Upper Paleolithic sites. Only two, Jarama VI and Zafarraya, were found to contain material that could be reliably dated. In both sites, Middle Paleolithic contexts were previously dated by radiocarbon to less than 42 ka calBP. Using ultrafiltration to purify faunal bone collagen before radiocarbon dating, we obtain ages at least 10 ka ¹⁴C years older, close to or beyond the limit of the radiocarbon method for the Mousterian at Jarama VI and Neanderthal fossils at Zafarraya. Unless rigorous pretreatment protocols have been used, radiocarbon dates should be assumed to be inaccurate until proven otherwise in this region. Evidence for the late survival of Neanderthals in southern Iberia is limited to one possible site, Cueva Antón, and alternative models of human occupation of the region should be considered.