Like elsewhere in Australia, the archaeology of northwest Queensland has focused on the antiquity of occupation and the continuity of that occupation through the Last Glacial Maximum (LGM), in an attempt to better understand the adaptive capabilities and strategies of early humans. Veth (1989, 1993) has hypothesised that the northwest Queensland savannah, as an important ‘corridor’ for the colonisation of Australia (e.g. Bird et al. 2005; Horton 1981), should contain ‘early’ sites; and furthermore that with the climatic deterioration associated with the LGM, such sites should fit one of two patterns: (1) they will be abandoned and display a cultural hiatus; or, (2) if located in resource-rich zones within catchments (‘local refuges’), they will continue to be utilised, though subsistence strategies will be modified to rely more heavily on locally available resources. The northwest Queensland sites of Colless Creek at Lawn Hill (Hiscock 1984, 1988), and GRE8 near Riversleigh (Slack 2007:218-251; Slack et al. 2004), both fit the second pattern, i.e. persistent occupation through the LGM with altered strategies to cope with increased aridity. However, outside these local refugia, sites pre-dating the LGM have not yet been located in the northwest Queensland savannah. For example, Mickey Springs 34 (Porcupine Gorge) provides evidence for human occupation from c.10,000 BP (Morwood 1990, 1992, 2002) and Cuckadoo Shelter in the Selwyn Ranges (Davidson et al. 1993) provides a near basal date of 15,270}210 BP; Veth (1989:87) argued that such sites reflect the post-LGM expansion of groups from refuges. The evidence available to date raises the question as to whether the wider northwest Queensland savannah corridor was indeed occupied in the pre-LGM period, when rainfall levels were higher and there was greater availability of surface water and food resources (cf. Hiscock and Wallis 2005).