Nitrogen oxide (NO<sub>x</sub>=NO+NO<sub>2</sub>) emissions from various sources contribute to the ozone budget. The identification of these contributions is important, e.g. for the assessment of emissions from traffic. The non-linear character of ozone chemistry complicates the online diagnosis during multi-decadal chemistry-climate simulations. A methodology is suggested, which is efficient enough to be incorporated in multi-decadal simulations. Eight types of NO<sub>x</sub> emissions are included in the model. For each a NO<sub>y</sub> (=all N components, except N<sub>2</sub> and N<sub>2</sub>O) tracer and an ozone tracer is included in the model, which experience the same emissions and loss processes like the online chemistry fields. To calculate the ozone changes caused by an individual NO<sub>x</sub> emission, the assumption is made that the NO<sub>x</sub> contributions from various sources are identical to the NO<sub>y</sub> contributions. To evaluate this method each NO<sub>x</sub> emission has been increased by 5% and a detailed error analysis is given. In the regions of the main impact of individual sources the potential error of the calculated contribution is significantly smaller than the contribution. Moreover, the changes caused by an increase of the emissions of 5% were detected with a higher accuracy than the potential errror of the absolut contribution.