Crop yield maps may contain substantial corollary information regarding the distribution of yield related soil properties across a landscape. One of these properties is water holding capacity (WHC). Since WHC is an important parameter for crop models and is also critical for crop yield, our objective was to determine if WHC could be estimated by matching simulated yield with yield map data. We collected soil cores for water retention measurements and recorded plant phenological stages from 60 plots on four transects over two growing seasons (1997 and 1998). Soil cores were also sampled on 40 other locations set out on a grid pattern. We utilized a simple water budget model that uses the relative transpiration ratio to calculate relative yield from available water in the soil pro®le. Rainfall, potential evapotranspiration and soil water holding capacity are input. An optimization program varies the WHC to produce a grain yield similar to the one from the yield map at a particular location. This analysis was carried out over several scales by averaging yields over 55 m  71 m, 27 m  35 m, and 11 m  14 m areas. Yield data from 2 years were used. Yields from the transects in both years were signi®cantly related to measured WHC in the surface 0±10 cm of soil. The calculated stress indices from the water budget model and estimated available WHC calculated for the 1997 data were similar to those calculated for the 1998 data where data were aggregated in 27 by 35 m or larger blocks. The contour map of estimated WHC was similar to the map of measured WHC for some features though there were also some differences. Use of multiple years of yield data are required to give stable results for estimated water holding capacities.. This information could be used in a farm management plan by allowing a producer to classify a ®eld into areas that are buffered against drought and areas more susceptible to drought. Published by Elsevier Science B.V.