Archaeological excavation over 30 years ago in a small granite cave on Great Glennie Island, 7 km off the southernmost tip of mainland Australia, revealed five stratigraphically distinct midden units with intermittent occupation over the last 2000 years. Fluctuations in tool stone (quartz and flint) abundance and the nature of tool use track a shift in economic focus from one dominant taxon (seabird) during initial visits to four (seal, seabird, fish and shellfish) in the last 500 years. An analysis of flint cortex distributions indicates that during the second visit people carried core tools and retouched flint flakes, and during later visits they transported unworked flint nodules for knapping flakes as needed on the island. These data have been integrated with a functional analysis of all the stone artefacts to evaluate a model of human settlement based on island biogeography formulated by Jones (1976). I explore whether island toolkits can be linked with shifts in food procurement and argue that the island was visited infrequently because there was no compelling need to regularly harvest its rich food resources.