For crop simulation models used in global analysis the most illusive and yet important model inputs are climate data. But there is a wide range of quality and derivation in available global climate datasets. Additionally, there is a trade-off between temporal and geospatial resolution inherent in deriving climate datasets. Good crop models typically operate on a daily time-step, but daily climate data are only available at specific observation points or interpolated from these points at low geospatial resolutions.  Alternative derivation methods achieve high geospatial resolution, but offer only monthly or temporally longer averaged climate data time-steps.  In this study, the range of climate dataset options are analyzed and compared in crop analysis of Yp at 12 sites in China, the US and Germany for rice, maize, and wheat respectively.  Preliminary results indicate that observed data should be used wherever possible and temporal resolution is more significant than geospatial resolution.  More accurate estimations of yields in crop models will lead to more meaningful and accurate recommendations for major stakeholders.