In order to estimate the spatial variation within well-defined study areas, nitrogen dioxide was measured with diffusion samplers (Palmes tube) in 40-42 sites each in Germany (Munich), the Netherlands and Sweden (Stockholm County). Each site was measured over four 2-week periods during 1 year (spring 1999 to summer 2000). In each country, one reference site was measured during all periods and the results were used to adjust for seasonal variability, to improve the estimates of the annual average. Comparisons between the chemiluminescence method (European reference method) and Palmes tube measurement indicated a good agreement in Germany (with a ratio of 1.0 for Palmes tube/chemiluminescence) but underestimation for Palmes tube measurement in the Netherlands and Sweden (0.8 for both countries). The r2 values were between 0.86 and 0.90 for all three countries. The annual average values for NO2 for different sampling sites were between 15.9 and 50.6 (mean 28.8 microg/m3) in Germany, between 12.1 and 50.8 (mean 28.9 microg/m3) in the Netherlands and between 6.1 and 44.7 (mean 18.5 microg/m3) in Sweden. Comparing spatial variation between similar sites in the three countries, we did not find any significant differences between annual average levels for urban traffic sites. In Sweden, annual average levels in urban background and suburban backgrounds sites were about 8 microg/m3 lower than comparable sites in Germany and the Netherlands. Comparing site types within each country only urban traffic sites and suburban background sites differed in Germany. In the Netherlands and Sweden, the urban traffic sites differed from all other sites and in Sweden also the urban background sites differed from the other background sites. The observed contribution from local traffic was similar in the Netherlands and Sweden (10 and 8 microg/m3, corresponding to 26-27% of the NO2 concentration found in the urban traffic sites). In Germany, the contribution from local traffic was only 3 microg/m3, corresponding to 9% of the NO2 concentration found in the urban traffic sites. The spatial variation was substantially larger for NO2 than the variation for PM2.5 and similar to PM2.5 absorbance, measured in the same locations.