Drought is a major cause of limited productivity in agroecosystems and affects our ability to produce adequate food, fuel, and fiber.  Existing methods that predict the magnitude and spatial extent of drought risk are typically based on long-term precipitation and temperature data. We hypothesize that in situ monitoring of soil moisture can be used to create drought risk assessments with increased accuracy.  However, long-term soil moisture data are not widely available and effective surrogates are needed. We evaluated differences between 1) a drought risk assessment based on soil moisture data from the Oklahoma Mesonet, an automated network of 120 automated meteorological stations across Oklahoma, and 2) a drought risk assessment based on widely available long-term weather data (Purcell et al., 2003).  Using a 13-yr data set for eight locations, the probability of having a soil water deficit (SWD) sufficient to cause plant water stress was estimated for each day of the year (DOY) by these two methods. For method one, a site-specific threshold SWD was defined as depletion of 50% of the total available water, whereas for method two a uniform SWD threshold of 50 mm was used.  The methods agreed well for predicting the first and last DOY with >20% drought probability at these locations and for predicting the DOY with highest probability of drought.  Results also revealed a significant and high correlation between the methods' predictions of drought risk for each DOY, with r² varying from 85.2 to 99.8%.  This is the first known validation of the Purcell et al. (2003) drought risk assessment method using measured soil moisture data.