Homo erectus was the first human lineage to disperse widely throughout the Old World, the only hominin in Asia through much of the Pleistocene, and was likely ancestral to H. sapiens. The demise of this taxon remains obscure because of uncertainties regarding the geological age of its youngest populations. In 1996, some of us co-published electron spin resonance (ESR) and uranium series (U-series) results indicating an age as young as 35-50 ka for the late H. erectus sites of Ngandong and Sambungmacan and the faunal site of Jigar (Indonesia). If correct, these ages favor an African origin for recent humans who would overlap with H. erectus in time and space. Here, we report 40Ar/39Ar incremental heating analyses and new ESR/U-series age estimates from the "20 m terrace" at Ngandong and Jigar. Both data sets are internally consistent and provide no evidence for reworking, yet they are inconsistent with one another. The 40Ar/39Ar analyses give an average age of 546±12 ka (sd±5 se) for both sites, the first reliable radiometric indications of a middle Pleistocene component for the terrace. Given the technical accuracy and consistency of the analyses, the argon ages represent either the actual age or the maximum age for the terrace and are significantly older than previous estimates. Most of the ESR/U-series results are older as well, but the oldest that meets all modeling criteria is 143 ka+20/-17. Most samples indicated leaching of uranium and likely represent either the actual or the minimum age of the terrace. Given known sources of error, the U-series results could be consistent with a middle Pleistocene age. However, the ESR and 40Ar/39Ar ages preclude one another. Regardless, the age of the sites and hominins is at least bracketed between these estimates and is older than currently accepted. © 2011 Indriati et al.