Radiocarbon age control on the type Glenwood, Calumet, and Toleston shoreline features and on the abandoned Chicago outlet at the south end of the Lake Michigan basin provides a basis for reevaluating the timing and causes of high lake phases in the basin. Radiocarbon dates suggest that Glenwood-level (195 m) shoreline features formed between 14,100 and 12,700 yr B.P. (Glenwood I and II phases), Calumet-level (189 m) between 12,700 and 11,000 yr B.P. (Calumet I and II phases), and Toleston-level (184.5 m) between 5000 and 4000 yr B.P. (Nipissing phase), and that the Chicago outlet was cut to its present level (180 m) on bedrock while the lake was at the Glenwood level. This new chronology is inconsistent with J H. Bretz' hypothesis ((1951) American Journal of Science 249 , 401–429) that the progressive lowering of lake level resulted from episodic down-cutting of the outlet. Instead, the changes in lake level appear to relate to changes in the amount of glacial meltwater and precipitation entering the basin. We hypothesize that the Glenwood phases correspond with times when discharge from the Huron and Erie basins also entered the Lake Michigan basin (Lake Border and early Port Huron glacial phases), the Calumet phases with times when drainage was from the Lake Michigan basin alone (late Port Huron and Two Rivers glacial phases), and the Nipissing phase with the postglacial middle Holocene transgression caused by differential uplift in the basin. Estimates of relative net inputs to the basin during the Glenwood, Calumet, and Nipissing lake phases are consistent with estimates of relative outputs (i.e., discharge through the Chicago outlet); the magnitude of relative differences in inputs and outputs between phases is sufficient to explain lake-level changes of 4.5 to 6 m.