A network of Swedish sites, ranging from regional rural background, urban background to kerbside sites, measuring PM2.5 and PM10 on one hour basis has been operated over 2 years in Sweden. The results show two important features determining the concentrations, long distance transport of particles and re-suspension of road dust. Strong low-level inversions especially in the inland of Northern Sweden during winter lock out the long distant transport and lead to less dilution of the local emissions and induce at times very high concentrations. Typical annual mean values for PM2.5 at rural background, urban background and kerb side sites are 5 - 10, 6 - 11 and 10 - 15 ?g/m³, respectively, while for PM10 it is 7 - 13, 12 - 18 and 20 - 30. The ratio PM2.5/ PM10 is about 0.8, 0.6 - 0.7 and 0.4 - 0.6 at the rural background, urban background and kerb side sites respectively. The strongly decreasing ratio implies the strength of road dust emissions and other mechanical wearing processes giving coarse particle emissions. Using NOx as a tracer for traffic emission and assuming a fixed relation to NOx for the particle exhaust emissions the non-exhaust emission factors for PM2.5 and PM10 is estimated to 25 and 200 mg/vehkm compared to the estimated 23 mg/vehkm for exhaust related particle emissions. The ranges observed for the non-exhaust emission factors for PM2.5 and PM10 are 0 - 76 and 25 - 554 mg/vehkm, respectively. Totally the non-exhaust emission factor for PM10 is 9 times larger than the PM10 exhaust emission factor but for PM2.5 the non-exhaust emission factor is equal to the exhaust emission factor. The high correlation between PM10 and PM2.5 found in this study are explained in this study and possibly others too by the emission of road dust and other coarse particle sources are closely linked to traffic intensity and thus primary exhaust, together totally dominating PM10. In background situations PM2.5 is as such dominating PM10 and thus explaining the correlation. Road dust emerges as the major local source responsible for exceedance of limit values in Sweden. However it is superimposed on strongly enhanced background concentrations due to long distance transport. It is obvious from this study that PM2.5 has a strong component of non-exhaust coarse particle sources. PM2.5 thus do not reflect the correct concentrations due to combustion sources and long range transported emissions. Using PM2.5 will complicate the assessment of sources and the choice of abatement strategy. Use of PM1 would avoid these problems.