Soil water balance models are commonly applied to monitor the availability of root-zone water for plant uptake of water and the occurrence of agricultural drought. This application is based on the assumption that temporal anomalies in root-zone water content provide a detectable precursor to the onset of wide-scale vegetation water stress and associated declines in agricultural productivity. However, such predictability can only be leveraged when root-zone soil water anomalies are monitored with sufficient precision to detect (potentially modest) levels of correlation between current soil moisture and future vegetation conditions. While a variety of model physics and forcing data sets have been proposed for global water balance modeling, relatively little objective benchmarking has occurred to evaluate the relative merits of various approaches for monitoring agricultural drought. This presentation will present a potential methodology for performing such benchmarking. The approach is based on comparing lagged anomaly correlations between root-zone soil moisture output from multiple land surface models (with contrasting model physics and forcing data) and visible/near-infrared vegetation indices like the Enhanced Vegetation Index (EVI). Within water-limited areas, improved precision in root-zone soil moisture predictions yields higher-lagged correlations with future EVI anomalies. In this way, various model physics and forcing data sets will be objectively evaluated based on their ability to contribute useful information to agricultural drought forecasts. Implications for operational globally drought monitoring effects will be discussed and, for a sub-set of the models considered, the added impact of assimilating satellite-based surface soil moisture retrievals will be considered.