Abstract A crop simulation model must first be capable of representing the actual performance of crops grown in any particular region before it can be applied to the prediction of climate change impacts. Erosion productivity impact calculator (EPIC) simulations of crop productivity in the Missouri-Iowa-Nebraska-Kansas (MINK) region under the 1951–1980 climate were compared with the US Department of Agriculture's ‘County Yield Estimates’ data (averaged over 1984–1987), with expert estimates of yields for each ‘representative’ farm and with the results of agronomic experiments reported in the literature. Most EPIC-simulated yields agreed to within ±20% with USDA reported yields and expert estimates, although there were some outliers. EPIC-simulated yields, evapotranspiration and water use efficiency fell well within the range of experimental results. Perfect agreement with observed crop performance was not a requisite nor should it have been expected. We judged the EPIC simulations sufficiently reliable to justify use of the model in simulating the effects of climate change on crops.