The warm-season rainfall variability over the US Great Plains is examined in a climate simulation and compared with observations to explore the role of land-atmosphere interaction in the model. The results show that compared to observations, the model simulations underestimate summertime rainfall by more than 12% but overestimate its standard deviation by 25--54%, with the larger value referring to interannual variability. Linear regression shows that the rainfall variability is connected with evapotranspiration (ET) anomalies, but mostly to evaporation rather than transpiration. Since the evaporation represents an immediate response of the land surface to atmospheric conditions but the transpiration reflects soil moisture memory with considerably longer time scales, the realism of the correlated variability between separate ET components and precipitation is an important aspect of how climate models realistically address land-atmosphere interaction.