Several lines of evidence suggest that forest growth in many regions is declining as a consequence of changing climate. To predict the fate of forests in the future, a quantitative understanding of how the key climate variables (insolation, precipitation and temperature) interact with forests to cause the decline is a pressing need. Here we use a regionally-averaged tree-ring width index (RWIr ) to quantify forest growth in the Southwest United States (SWUS). We show that over a period of 100 years, SWUS RWIr bifurcated into forest stands with enhanced (healthy) and reduced (declining) branches when regressed on shortwave-radiation and temperature, respectively. The reduced branch was controlled overwhelmingly by drought as measured with a regionally-averaged precipitationevapotranspiration index (SPEIr ). As SPEIr approached -1.6 (previously shown as a tipping-point for SWUS conifer forest growth), RWIr approached zero and in extreme drought years, wide spread tree mortality has been observed. Modeled trends in SPEI based on four IPCC-GHG scenarios predict SWUS SPEIr falling below -1.6 more or less continuously within a few decades. With drought expanding north- and eastward over larger areas, tree mortality may become a semi-continental phenomenon with coniferous forests transitioning to more xeric ecosystems. Our results provide insights into how to differentiate functions of climate impacts on forest growth and how to identify tipping-point control parameters for forest regime transitions.