Agriculture and land use change for agriculture account for nearly a quarter of total global greenhouse gas emissions, with emissions coming from both on-farm management practices and land conversion for agricultural use. Agricultural practices that increase yields typically reduce land conversion, while reductions in yields can lead to additional conversion globally. Because of these two mechanisms, a complete understanding of the net greenhouse gas impacts of agricultural practices requires quantifying both direct emissions and impacts on yield and land use change. Here, we assess the net greenhouse gas impacts of US irrigation by first quantifying the productivity benefits of irrigation, then using a global economic model to project the land use change that would occur without the additional production, and finally using high-resolution estimates of carbon stocks to estimate the greenhouse gas impacts of the avoided land use change attributable to irrigation. We estimate that US irrigation avoids 6.2 Gt CO₂e of land use change emissions, equal to >300 years’ worth of direct annual emissions from irrigation. The modeling framework developed in this study is translatable to other agricultural management practices and can be used to comprehensively assess direct and indirect emissions.