AbstractFuture crop yield increases are unlikely to keep pace with the dietary needs of a global human population expected to reach nine billion by 2050. This study used United States Department of Agriculture county‐level yield data and autoregressive moving‐average models to examine how changes in maize (Zea mays L.), soybean (Glycine max L.), and winter wheat (Triticum aestivum L.) yields, temporal variability in yields, and yield gaps have varied across space and time from 1970–2017. The majority of county‐level yields have increased linearly, although the increases in wheat lag behind corn and soybean. Where trends were nonlinear, accelerating yields were found in more mesic regions east of the Great Plains, and decelerating yields were found in the drier central and western United States. Mean crop yields were positively correlated with rate of yield increase and negatively correlated with interannual variability. Hotspots were identified in Minnesota, Iowa, Illinois, Nebraska, and some West Coast states where crop yields are currently the largest, have the lowest yield gaps, and since 1970, have had the highest rates of change and/or are experiencing an acceleration of annual yield gains. Across all crop types, the counties with the lowest average yields, highest yield gaps, lowest rates of yield increase over time, and/or deceleration in yield increases were predominantly found in the central United States, including the Dakotas, Kansas, Oklahoma, and Texas. Regions of greatest performance generally have fertile soils, plentiful growing season rainfall, and optimal growing season length and temperatures, or are benefitting from irrigation.