Field studies were conducted in 1986, 1987, 1988, and 1989 to determine the stability of crop loss functions across site by year environments. Environment was a significant source of variation for the soybean crop loss function as influenced by velvetleaf, but not as influenced by tall waterhemp and common sunflower. Weed density was a highly significant source of variation for all weed species studied. Regressions between percent soybean seed yield reductions and weed populations were linear. The velvetleaf interference regression equations were divided into two groups, those with high soybean-yielding intercepts and those with low-yielding intercepts, to explain the variance observed across environments. The regression equation for the high-yielding intercept group was Ŷ = 4.24X while the low-yielding group was Ŷ = 2.14X, where Y is percent soybean yield reduction and X is weed density per 10.7 m of soybean row. Tall waterhemp and common sunflower regression equations were determined to be Ŷ = 1.37X and Ŷ = 6.52X, respectively. Confidence intervals were used to account for the variance associated with the mean regression equation for each model and to develop economic threshold models that include risk aversion principles.