Concentrations of NO 2 , NO, and O 3 from a rooftop monitoring station in Gothenburg, Sweden (2002 were analysed to characterise NO 2 pollution. [NO 2 ] was shown to correlate strongly and non-linearly with [NO x ] (NO x =NO+NO 2 ), in line with observations in other cities. The [NO 2 ] to [NO x ] fraction fell initially with increasing [NO x ]. At [NO x ] levels >200 ppb, the decline in [NO 2 ]/[NO x ] with increasing [NO x ] levelled out and [NO 2 ]/[NO x ] converged towards approximately 0.15-0.16, independent of [NO x ]. Data from a traffic route site showed the same pattern. This value of [NO 2 ]/[NO x ] at high [NO x ] can be interpreted as the NO 2 fraction of the NO x emissions from vehicle exhaust. Situations with high NO x pollution and minimum [NO 2 ]/[NO x ] were always associated with [O 3 ] close to zero. Plotting [Ox] (Ox=NO 2 +O 3 ) vs. [NO x ] provided a strong linear correlation for situations dominated by local pollution ([NO]/[NO 2 ]>1). The slope of the regression, a measure of the primary NO 2 fraction in NO x emissions, was 0.13 during the day and 0.14 during the night. With stronger winds, the rooftop monitoring station became more similar, in terms of NO 2 pollution, to a city street site and a traffic route site, although [NO 2 ] was almost always higher at the street/traffic route locations. The EU standard for the annual average of [NO 2 ] (40 μg m −3 ) was exceeded, while the hourly standard (200 μg m −3 , not to be exceeded more than 18 times per year by 2010) was not exceeded at any of the sites.