The spatial and temporal performance of an ensemble of five gridded climate datasets (precipitation) (North American Regional Reanalysis, European Centre for Medium-Range Weather Forecasts interim reanalysis, European Union Water and Global Change (WATCH) Watch Forcing data ERA-Interim, Global Forcing Data-Hydro, and The Australian National University spline interpolation) was evaluated towards quantifying gridded precipitation data ensemble uncertainty for hydrologic model input. Performance was evaluated over the Nelson–Churchill Watershed via comparison to two ground-based climate station datasets for year-to-year and season-to-season periods (1981–2010) at three spatial discretizations: distributed, sub-basin aggregation, and full watershed aggregation. All gridded datasets showed spatial performance variations, most notably in year-to-year total precipitation bias. Absolute minimum and maximum realizations were generated and assumed to represent total possible uncertainty bounds of the ensemble. Analyses showed that high magnitude precipitation events were often outside the uncertainty envelope; some increase in spatial aggregation, however, enveloped more observations. Results suggest that hydrologic models can reduce input uncertainty with some spatial aggregation, but begin to lose information as aggregation increases. Uncertainty bounds also revealed periods of elevated uncertainty. Assessing input ensemble bounds can be used to include high and low uncertainty periods in hydrologic model calibration and validation.