Future hydroclimate change is expected to generally follow a wet-get-wetter, dry-get-drier (WWDD) pattern, yet key uncertainties remain regionally and over land. It has been previously hypothesized that lake levels of the Last Glacial Maximum (LGM) could map a reverse analog to future hydroclimate changes due to reduction of CO 2 levels at this time. Potential complications to this approach include, however, the confounding effects of factors such as the Laurentide Ice Sheet and lake evaporation changes. Using the ensemble output of six coupled climate models, lake energy and water balance models, an atmospheric moisture budget analysis, and additional CO 2 sensitivity experiments, we assess the effectiveness of the LGM as a reverse analog for future hydroclimate changes for a transect from the drylands of North America to southern South America. The model ensemble successfully simulates the general pattern of lower tropical lake levels and higher extratropical lake levels at LGM, matching 82% of the lake proxy records. The greatest model-data mismatch occurs in tropical and extratropical South America, potentially as a result of underestimated changes in temperature and surface evaporation. Thermodynamic processes of the mean circulation best explain the direction of lake changes observed in the proxy record, particularly in the tropics and Pacific coasts of the extratropics, and produce a WWDD pattern. CO 2 forcing alone cannot account for LGM lake level changes, however, as the enhanced cooling from the Laurentide ice sheet appears necessary to generate LGM dry anomalies in the tropics and to deepen anomalies in the extratropics. LGM performance as a reverse analog is regionally dependent as anti-correlation between LGM and future P − E is not uniformly observed across the study domain.