Several factors, including multi-year drought, declining aquifer levels, and new water regulations, are contributing to reduced availability of irrigation water in the semi-arid area of west-central Nebraska. Since many farmers in this area do not have enough water to meet the seasonal water requirements of crops like corn and soybean, maximizing yield produced per unit of water under deficit irrigation conditions is becoming increasingly important. This study was conducted to quantify the grain yield response of soybean [Glycine max (L.) Merr.] to deficit irrigation, and to determine which seasonal water variables correlated best to soybean grain yield under deficit irrigation. The study was conducted during 2002 at Curtis, and 2003 and 2004 at North Platte, Nebraska. Nine deficit irrigation treatments, including different irrigation amounts and timings, were studied in 2002 and 2003, and eight treatments were studied in 2004. Soybean grain yields across years and sites were best related to the seasonal ratio of the actual crop evapotranspiration and the crop evapotranspiration when soil water was not limiting (ETd/ETw), and to the seasonal ratio of actual crop transpiration and crop transpiration when soil water was not limiting (Td/Tw). Both of these seasonal ratios were linearly related to grain yield with R2 = 0.91 when combining data for all seasons. The crop water productivity (CWP) (yield per unit of seasonal ETd) linearly increased with both ETd/ETw (R2 = 0.72) and Td/Tw (R2 = 0.72), but was best correlated to the daily positive difference between the actual and the theoretical fraction of total available soil water in the root zone that can be depleted before crop water stress occurred, accumulated for the entire season (seasonal pdiff) (R2 = 0.77). A linear relationship between the cumulative ETw and fraction of season (function of days after emergence) was found. This relationship developed for a given location could be used to extrapolate seasonal ETw for in-season irrigation management. Poor correlation was found between CWP and other variables such as total irrigation, rain + irrigation, and total water. The results of this study can provide useful information for soybean irrigators to make better management decisions under deficit irrigation conditions.