Despite the fact that it is the total crop production that shapes future food supply rather than one of its single component, previous studies have mainly focused on the changes in crop yield. It is possible that recent gains in crop production are mainly due to improvement of yield rather than growth of harvest area. However, it remains unclear about the geographical patterns of their relative contributions at fine scales and the possible mechanisms. Analysis of US maize production shows that maize production has increased significantly at a rate of 2.1%/year during 1980– 2010. Although yield is the dominant factor contributing to production growth for the country as a whole, the importance of harvest area has become more evident with time. In 56% of US’s maize growing counties, harvest area has also contributed more than yield to production changes. High spatial correlation between the change rates of harvest area and production is observed (R = 0.96), while a weak relation (R = 0.21) is found between the spatial patterns of yield and production. This suggests that harvest area has exerted the dominant role in modulating the spatial distribution pattern of maize production changes. Further analysis suggests that yield and harvest area respond differently to climate variability, which has great implications for adaptation strategies. Comparing 11 state-of-the-art crop model simulations against census data reveals large bias in the simulated spatial patterns of maize production. Nevertheless, such bias can be reduced substantially by incorporating the observed dynamics of harvest area, pointing to a potential pathway for future model improvement. This study highlights the importance of accounting for harvest area dynamics in assessing agricultural production empirically or with crop models.