Exposure of crops to high concentrations of ozone can cause substantial reductions in yield that pose a serious threat to global food security. Here we provide comprehensive estimates of yield losses for key crops across China between 2014 and 2017 attributed to ozone using a number of new approaches. We use an air quality model at 5 km resolution and crop-specific dose-response functions developed for both concentration- and flux-based metrics. We bias correct modelled ozone concentrations and metrics using observations from more than 1000 locations. We find that on a 4-year average basis, production losses of key crops are 34–91 million metric tonnes (Mt/yr), dependent on the approach used, with highest losses in Henan province. At a national level, loss of winter wheat production derived using a China-specific dose-response function increased by 82% from 2014 to 2017, with large interannual variations in the North China Plain and in eastern China. Winter wheat losses estimated using flux-based functions, which require robust simulation of stomatal conductance and underlying vegetation physiology, are significantly lower, at 30 Mt/yr. We show that the definition of the growing season may have a greater impact on estimated losses than small biases in ozone surface concentrations. Although uncertainties remain, our findings demonstrate that increasing ozone concentrations have substantial adverse impacts on crop yields and threaten food security in China. It is important to control ozone concentrations to mitigate these negative impacts.