Water is a key variable driving the composition and productivity of pastures and rangelands, and many of the ecosystems in these grasslands are highly sensitive to changes in water supply. The possibility that elevated CO2 concentrations may alter plant water relations is therefore particularly relevant to pastures and rangelands, and may have important consequences for grassland ecosystem function, water use, carbon storage and nutrient cycling. The planning of effective research to better understand these changes requires attention to both: (i) gaps in knowledge about CO2 and water interactions, and (ii) knowledge of how precisely the effects of CO2 must be understood in relation to other factors, in order to predict changes in grassland structure and production. A recent microcosm experiment illustrates that non-linear effects of CO2 and water stress could perturb primary production by triggering changes in grassland community composition. The magnitudes of the effects of CO2 on key grassland ecosystems remain to be precisely determined through ecosystem-level experiments. A simplified simulation of the impact of different levels of productivity change in a water-limited Australian rangeland system was conducted by varying effects of CO2 on radiation and water use efficiency. The results indicate that direct effects of CO2 may be moderated at the enterprise scale by accompanying changes in adaptive management by farmers. We conclude that future research should aim to construct quantitative relationships and identify thresholds of response for different grassland systems. The sensitivity of these systems to management (such as grazing pressure) should also be considered when developing integrated predictions of future effects of CO2 on water supply to grassland ecosystems.