Enhanced evapotranspiration (ET) over irrigated land and associated latent heat flux change can modify the climate. Model studies of such climate change effects of irrigation are commonly based on land use parameterizations, in terms of irrigated land area, or land area equipped for irrigation. Actual ET change, however, may also be driven by water use change in addition to land use change. This study quantifies and compares ET changes due to changes in climate, land use, and water use from the preirrigation period 1901–1955 to the recent period 1990–2000 (with irrigation) for the example case of Mahanadi River Basin (MRB) in India. The results show that actual water use per unit area of irrigated land may vary greatly over a hydrological drainage basin. In MRB, much higher water use per irrigated land unit in the downstream humid basin parts leads to higher vapor flux by ET, and irrigation-induced ET flux change, than in the upstream, water-stressed basin parts. This is consistent with water supply limitations in water-stressed basins. In contrast, the assumption in land use−based models that irrigation maintains high soil moisture contents can imply higher modeled water use and therefore also higher modeled ET fluxes under dry conditions than under humid conditions. The present results indicate water use as an important driver of regional climate change, in addition to land use and greenhouse gas-driven changes.