Downscaled climate data are available at fine spatial scales making them desirable to local climate change practitioners. However, without a description of their uncertainty, practitioners cannot know if they provide quality information. We pose that part of the foundation for the description of uncertainty is an assessment of the ability of the underlying climate model to represent the meteorological or weatherscale processes. Here, we demonstrate an assessment of precipitation processes for the Great Lakes region using the Bias Corrected and Spatially Downscaled (BCSD) Coupled Model Intercomparison Project phase 3 (CMIP3) projections. A major weakness of the underlying models is their inability to simulate the effects of the Great Lakes, which is an important issue for most global climate models. There is also uncertainty among the models in the timing of transition between dominant precipitation processes going from the warm to cool season and vice versa. In addition, warm-season convective precipitation processes very greatly among the models. From the assessment, we discuss how process-based uncertainties in the models are inherited by the downscaled projections and how bias correction increases uncertainty in cases where precipitation processes are not well represented. Implications of these findings are presented for three regional examples: lake-effect snow, the spring seasonal transition, and summertime lake-effect precipitation.