The recent increase in dune studies in North America has been heavily focused in the Great Plains, while less attention has historically been given to the dune fields east of the Mississippi River. Here we report ages and suggest a potential sediment source for sand dunes in the Green River Lowland, Illinois, which may provide a better understanding of the dynamic interactions between eolian, glacial, lacustrine and fluvial processes that shaped the landscapes of the upper Midwest. Seven coherent optically stimulated luminescence ages (OSL, or optical ages) obtained from four sites suggest that major dune construction in the Green River Lowland occurred within a narrow time window around 17,500 ago. This implies either an enhanced aridity or an episodic increase of sediment supply at 17,500 years ago, or combination of the both. Contrary to previous assertions that dune sand was sourced from the deflation of the underlying outwash sand deposited when the Lake Michigan Lobe retreated from the area, we propose that Green River Lowland dunes sand originated from the Green Bay Lobe through the Rock River. Specifically, sediment supply increased in the Rock River valley during drainage of Glacial Lake Scuppernong, which formed between ???18,000 and 17,000 years ago, when the Green Bay Lobe retreated from its terminal moraine. The lake drained catastrophically through the Rock River valley, providing glacial sediment and water to erode the preexisting sandy sediments. Throughout the remainder of the late Pleistocene, the Laurentide Ice Sheet drained into larger more northerly glacial lakes that in turn drained through other river valleys. Therefore, the dunes in the Green River Lowland formed only during the catastrophic drainage of Glacial Lake Scuppernong, but were stabilized through the remainder of the Pleistocene. This scenario explains the abrupt dune construction around 17,500 years ago, and explains the lack of later dune activity up to the Pleistocene-Holocene transition. OSL and radiocarbon ages also indicate that dunes were reactivated during the early, middle and late Holocene. Some eolian activation occurred within well-defined dry intervals in the upper Midwest, suggesting that increased aridity may have been the primary driver in mobilizing sand. However, many ages do not correspond to drier periods. In contrast to the relative coherency of the Pleistocene OSL ages from multiple study sites, the Holocene OSL ages do not overlap from one site to another, suggesting that increased aridity alone cannot explain the multiple phases of dune reactivation in the Holocene. Therefore, we conclude that the combined effect of localized disturbances and greater aridity acted in concert to increase eolian sand activity in the Holocene. The multiple periods of eolian activity during the Holocene suggest a high potential for future sand activation in the region, and these results are informative for environmental prediction and potential future mitigation.