Knowledge of Earth’s climate history and sensitivity, combined with modeling past and future climate, are central to informing policy decisions regarding future climate change. The hydrologic response to future warming scenarios due to increased anthropogenic CO2 emissions remains uncertain. Freshwater availability in the arid western United States is projected to decrease in availability as increased agricultural, urban and industrial uses continue to stress supplies. Motivated by the potential for dramatic future hydrologic changes, studies recording the abrupt transitions between different equilibrium states of natural past climate variability shed light on our understanding of the modern climate system.The presence of pluvial lakes in the Basin and Range Province, in the western United States, during the late Pleistocene (40 to 10 ka) indicates far greater moisture availability during the Pleistocene glacials. This study investigates the timing and magnitude of the most recent pluvial lake cycle that filled Surprise Valley, California using geophysical, geochemical and geochronologic tools. Spanning 31.2 to 4.6 ka, this new lake level record places the highest lake level, at 180 meters above present day playa, at 13.9 ± 1.2 ka. This age appears to be nearly synchronous with highstands of Lake Lahontan to the south and the Chewaucan Basin to the north. Additionally, most of the Basin and Range lake highstands, including Lake Surprise, follow peaks in precipitation minus evapotranspiration (P-ET) by 8-10 kyr. By compiling a diverse set of paleoclimate data available for western North America, I found that the timing and geographic distribution of lake highstands is inconsistent with increased precipitation in response to shifting westerly winds, the current model for the genesis of large lakes in western North America. Rather, lakes levels are more strongly correlated with changes in summer insolation, suggesting that lake highstands were likely facilitated by colder temperatures and increased humidity due to the presence of continental ice sheets and increased atmospheric convergence. I compared the constraints from lake and soil-based records to Atmosphere-Ocean General Circulation Model simulations from the Paleoclimate Model Intercomparison Project 2. Based on model-proxy intercomparison, the Atmosphere-Ocean General Circulation Models, the same models used to also assess future climatic changes, poorly predict hydrologic quantities for the Last Glacial Maximum.