For all of the 21^st century to date the American southwest has been in a megadrought that is amongst the worst of the last millennium. Here it is shown, using sea surface temperature (SST)-forced models and observations, that this has been driven by the cool tropics phase of the Pacific decadal oscillation that forced a decline in cool season precipitation. The summer soil moisture decline is also explained as a cross-season consequence of cool season precipitation decline. A large ensemble of model projections forced by realistic scenarios of Pacific and Atlantic Ocean decadal variability combined with radiatively-forced SST trends shows that even under the best-case scenario of a warm tropical Pacific and cool North Atlantic the southwest does not return to levels of water availability typical of the end of the last century. For almost all cases in the large ensemble, a decline in precipitation and increase in evapotranspiration in the cool season is shown to be critical for aridification. In the summer, soils are markedly drier, evapotranspiration decreases and vapor pressure deficit (VPD) increase. These adjustments in the land-atmosphere hydrology are consistent with the following causal flow: winter precipitation declines and winter evapotranspiration increases driven by higher VPD which reduces runoff and soil moisture. Spring and summer then inherit drier soils which leads to reduced evapotranspiration. As a result, atmospheric humidity cannot keep pace with rising warming-driven saturation humidity and VPD increases. The best-case scenario is one in which decadal ocean variability leads to winter precipitation increases that overwhelm the increase in evapotranspiration leading to soils that are less dry or wetter in spring. In this case higher VPD in spring drives higher evapotranspiration. Soil moisture changes in summer in the best-case scenario are not robust. In the worst-case (cool tropical Pacific and warm North Atlantic), spring soils are so dry that they drive evapotranspiration and surface air humidity to decline and VPD is even more positive. The expectation that atmospheric warming reduces soil moisture by enhancing evapotranspiration is valid but with the important caveat that this works in the cooler seasons only. The work draws attention to the important control that cool season precipitation exerts in the land-atmosphere coupled system of the southwest.