The first case of glyphosate resistant weeds was documented in 1998, two years after the commercialization of genetically engineered, herbicide tolerant (HT) corn and soybeans. Currently, over 14 resistant weed species affect U.S. crop production areas. This study uses a novel two-stage, structural model to derive cost functions for pre-emergent and postemergent herbicide use. The parameters of the structural model are estimated using censored nonlinear full information maximum likelihood. Control functions from firststage regressions account for endogeneity of farmers’ profit-maximizing decisions. The results of the analysis are used to determine the extent to which resistance has eroded the benefits associated with glyphosate applications in HT production systems. Our preliminary results suggest that glyphosate resistance may have decreased weed control by as much as 11 percentage points from 2005 to 2010.