AbstractAgricultural technology has increased crop yield potentials, but on rain‐fed crops yields are still severely reduced with the normal climatic frequency of drought. Objectives were (1) to determine an interaction regression of county average corn (Zea mays L.) yield on a soil moisture stress variable and technology trend and (2) to estimate the probability of soil moisture stress and resulting average corn yield in Tippecanoe County, Indiana. The soil moisture stress variable (Sc) was the sum of modeled daily ratios of actual to potential evapotranspiration [Σ(ET/PET)] over critical corn growth and development periods. The interaction regression model of corn yield on Sc andyield on Sc andyield on Sc andyield on Sc and technology trend (T = year) for Tippecanoe County was associated with 70% of the variance in the 1961 to 1992 average county corn yields when Sc was a 90‐d period (S90) from 39 d before corn silking to 50 d after. With no moisture stress (S90 = 90), the technology trend over the last 32 yr was 0.17 t ha−1 yr−1 (2.7 bu acre−1 yr−1). With 1992 technology, each deficit unit of S90 reduced the yield 0.19 t ha−1 (3.1 bu acre−1). The distributions of S90 and predicted corn yield were highly negatively skewed. The probability of having an S90 less than 85 (at least some moisture stress), and a county corn yield less than 9.5 t ha−1 (152 bu acre−1) is 69%, but the probability of severe stress (S90 < 75) and corn yield less than 7.5 ± 0.8 t ha−1 (139 ± 13 bu acre−1) is 22%. For the same weather regime, the probability of moisture stress and resulting corn yields differs greatly for individual soils. For a poorly drained soil (Typic Argiaquoll) the probability of having an S90 less than 85 is 41%, but for a well‐drained soil (Typic Argiudoll) the probability is 90%.