Agricultural irrigation practices will likely be affected by climate change. In this paper, we use a statistical model relating observed water use by U.S. producers to the moisture deficit, and then use this statistical model to project climate changes impact on both the fraction of agricultural land irrigated and the irrigation rate (m³ ha⁻¹). Data on water withdrawals for US states (1985-2005) show that both quantities are highly positively correlated with moisture deficit (precipitation--PET). If current trends hold, climate change would increase agricultural demand for irrigation in 2090 by 4.5-21.9 million ha (B1 scenario demand: 4.5-8.7 million ha, A2 scenario demand: 9.1-21.9 million ha). Much of this new irrigated area would occur in states that currently have a wet climate and a small fraction of their agricultural land currently irrigated, posing a challenge to policymakers in states with less experience with strict regulation of agriculture water use. Moreover, most of this expansion will occur in states where current agricultural production has relatively low market value per hectare, which may make installation of irrigation uneconomical without significant changes in crops or practices by producers. Without significant increases in irrigation efficiency, climate change would also increase the average irrigation rate from 7,963 to 8,400-10,415 m³ ha⁻¹ (B1 rate: 8,400-9,145 m³ ha⁻¹, A2 rate: 9,380-10,415 m³ ha⁻¹). The irrigation rate will increase the most in states that already have dry climates and large irrigation rates, posing a challenge for water supply systems in these states. Accounting for both the increase in irrigated area and irrigation rate, total withdrawals might increase by 47.7-283.4 billion m³ (B1 withdrawal: 47.7-106.0 billion m³, A2 withdrawal: 117.4-283.4 billion m³). Increases in irrigation water-use efficiency, particularly by reducing the prevalence of surface irrigation, could eliminate the increase in total irrigation withdrawals in many states.