Downscaling methods are designed to improve the spatial detail and accuracy of information taken from coarser resolution climate model output. Despite rapid proliferation of studies and available methods, little is still known about the amount of uncertainty contributed by the many steps in a typical downscaling workflow. This chapter begins by describing key uncertainties inherent to dynamical and statistical downscaling, starting with the selection of the downscaling technique itself. Subsequent sources of uncertainty include choice of climate model and emissions scenarios driving the boundary conditions; area and location of the downscaling domain; spatial and temporal scaling methods; parameterization (or not) of sub-grid processes; predictor variable suite; and representation of climate variability. Ultimately, such decisions shape the physical credibility and uncertainty bounds of downscaled climate change projections. We then evaluate the extent to which intercomparison, and benchmarking studies can show the relative skill and added value of different downscaling techniques. We close by calling for greater application of downscaling in decision-making contexts and explain how this can be achieved despite myriad uncertainties. After all, this is the often-stated rationale for investing time and resources in downscaling in the first place.