We propose a new nonparametric approach for assessing future changes in annual stream flows in extreme drought years based on an ensemble of climate projections. We apply the method to the Potomac River basin, investigating whether future flows in the river may be impacted by “hot drought”, that is, increasing severity of hydrological drought caused by rising temperatures coupled with variability in precipitation. Long time series representative of annual climate in time periods of interest are constructed by pooling and concatenating shorter time series sampled from an ensemble of bias corrected and spatially downscaled climate projections, where the K-nearest neighbor method is used to select pool members. The pooled time series are of sufficient length to allow estimation of the probability distribution of a full range of future annual flows, including 1st percentile values, indicative of flow in an extreme drought year. An empirically derived climate response function for annual mean flow is used as this study’s simple hydrologic model. The resulting set of cumulative probability distributions can be used to compute scaling factors for future annual Potomac River flow which demonstrate the disparate impacts of climate change on high flow, average flow, and low flow years. For most scenarios considered, results indicate that though long-term mean precipitation and river flow will increase modestly in future years, annual flows in an extreme drought year will decrease. This new approach can provide multi-model consensus inputs for water supply planning models to support decision-making regarding new infrastructure for climate resilience.