Rainfall variability is a key driver of ecosystem structure and function in tallgrass prairie. Future rainfall patterns predicted by global climate models are expected to cause increasingly variable soil moisture availability, which should have substantial impact on net primary production and plant species composition. Temporal variability in growing season rainfall inputs has been experimentally increased (by 50%) beginning in 1998 in the Rainfall Manipulation Plot experiment at the Konza Prairie Biological Station in northeastern Kansas tallgrass prairie. After four growing seasons of manipulations, increased rainfall variability (with total quantity of rainfall held constant) led to lower and more variable soil water content (0-30 cm) and lower aboveground net primary production due to reduced biomass of subdominant warm-season C₄ grasses. In contrast, biomass, leaf net photosynthesis, and two measures of leaf water use efficiency in the dominant C₄ grass Andropogon gerardii showed no response to increased rainfall variability. In general these vegetation responses were equal or greater than those caused by imposing drought (by reducing rain event quantity by 30%) without altering the temporal variability in rainfall. We conclude 1) variability in rainfall inputs can have as much impact on soil moisture and net primary productivity as the quantity of rainfall inputs; 2) a key attribute of dominant plant species is their tolerance of variability in rainfall inputs; and 3) altered rainfall patterns may offset elevated CO2 impacts on soil moisture and productivity in grassland vegetation.