For areas in the United States not meeting the federal air quality standard for ozone, an issue of continuing controversy is the emphasis to be placed on controlling nitrogen oxides (NOx) in addition to emissions of reactive organic gases (ROG). To assess conditions under which ROG or NOx controls would be most effective, we have analyzed predictions from four studies that represent different locations and meteorological conditions, distinct chemical inputs, e.g., with or without significant biogenic emissions, and different air quality models. A consistent association is found between the sensitivity of ozone to reductions in ROG versus NOx emissions and the simulated total reactive nitrogen (NOy) at the time and place of peak ozone. In the studies examined, ozone was predicted to be reduced most effectively by ROG controls at locations where NOy concentrations exceeded a threshold value falling in the range of 10 to 25 ppb, whereas NOx controls were predicted to be more effective where NOy concentrations were below that threshold. The NOy level explains much of the difference in ozone sensitivity at different locations and provides a basis for comparison of predicted sensitivity from different models. In contrast, the morning concentration ratio of ROG to NOx that has been used in the past is a less reliable indicator of O3 sensitivity. Measurement of NOy concentrations along with ozone would assist in empirical testing of model predictions of responses to emission reductions.