This study estimates the non-stationary design flow in the Upper Green River Basin (UGRB), WYby applying a moving window with the streamflow projections under A2, A1B, and B1 scenarios.The reliable estimates of design flow are necessary for the water management and planning andthe operation of infrastructure to reduce the risk of economic loss, the environmental damage,and the loss of life.The design flow is generally estimated by the flood frequency analysis under the stationaryassumption, which the historical flood distribution is similar to the future flooddistribution. However, nature is not stationary. This study considers the non-stationarity ofthe future streamflow in determining the design flow.To estimate the design flow, this study uses the 112 bias-corrected spatially downscaled(BCSD) precipitation and temperature projections of the World Climate Research Programme (WCRP)Coupled Model Intercomparison Project phase 3 (CMIP3) datasets. Furthermore, this study usesthe Variable Infiltration Capacity (VIC) model to simulate the streamflow projections underthree climate scenarios (A2, A1B, and B1). The non-stationary design flows for each year usinga moving window are estimated for 2, 5, 10, 20, 30, 50, and 100 year return periods using theGeneral Extreme values distribution.The proposed method is applied to the Upper Green River Basin in Wyoming. Moreover, thisstudy presents the climate change impacts on the design flow.