In the United States, ensemble streamflow forecasts are generated from the Hydrologic Ensemble Forecast Service (HEFS), operated by the National Oceanic and Atmospheric Administration (NOAA) National Weather Service. On sub-seasonal to seasonal (S2S) timescales (i.e., weeks 3 - 4), HEFS is often driven by randomly resampled climatology, rather than a forecast. However, recent advances in ensemble weather prediction, including meteorological preprocessing techniques, present an opportunity to improve S2S streamflow predictions. Here, we present work that leverages two approaches developed at NOAA’s Physical Sciences Laboratory (PSL) that have been shown to improve S2S precipitation forecasts: (i) the parametric Censored Shifted Gamma Distribution method and (ii) the neural-network-based ResUnet, FiLM, and Climatological-Offramp. The improved weeks 3-4 probabilistic accumulated precipitation forecasts are used in a hydrological weighting (postprocessing) scheme, whereby the streamflow traces are resampled so as to reflect the precipitation forecast. As such, the streamflow postprocessing approach can incorporate the forecast, but in a way that can work seamlessly with existing operational workflows. The approach is demonstrated by applying the techniques to the Global Ensemble Forecast System v12 (GEFSv12) precipitation forecasts, but is general and could be applied to other S2S products. Applications will be discussed for case studies relevant to societal applications that would benefit from better S2S streamflow predictions, including Forecast Informed Reservoir Operations (FIRO).