In order to carry out efficient traffic and air quality management, validated models and PM emission estimates are needed. This paper compares current available emission factor estimates for PM10 and PM2.5 from emission databases and different emission models, and validates these against eight high quality street pollution measurements in Denmark, Sweden, Germany, Finland and Austria. The data sets show large variation of the PM concentration and emission factors with season and with location. Consistently at all roads the PM10 and PM2.5 emission factors are lower in the summer month than the rest of the year. For example, PM10 emission factors are in average 5-45% lower during the month 6-10 compared to the annual average. The range of observed total emission factors (including non-exhaust emissions) for the different sites during summer conditions are 80-130mgkm(-1) for PM10, 30-60mgkm(-1) for PM2.5 and 20-50mgkm(-1) for the exhaust emissions. We present two different strategies regarding modelling of PM emissions: (1) For Nordic conditions with strong seasonal variations due to studded tyres and the use of sand/salt as anti-skid treatment a time varying emission model is needed. An empirical model accounting for these Nordic conditions was previously developed in Sweden. (2) For other roads with a less pronounced seasonal variation (e.g. in Denmark, Germany, Austria) methods using a constant emission factor maybe appropriate. Two models are presented here. Further, we apply the different emission models to data sets outside the original countries. For example, we apply the "Swedish" model for two streets without studded tyre usage and the "German", model for Nordic data sets. The "Swedish" empirical model performs best for streets with studded tyre use, but was not able to improve the correlation versus measurements in comparison to using constant emission factors for the Danish side. The "German" method performed well for the streets without clear seasonal variation and reproduces the summer conditions for streets with pronounced seasonal variation. However, the seasonal variation of PM emission factors can be important even for countries not using studded tyres, e.g. in areas with cold weather and snow events using sand and de-icing materials. Here a constant emission factor probably will under-estimate the 90-percentiles and therefore a time varying emission model need to be used or developed for such areas. All emission factor models consistently indicate that a large part (about 50-85% depending on the location) of the total PM10 emissions originates from non-exhaust emissions. This implies that reduction measures for the exhaust part of the vehicle emissions will only have a limited effect on ambient PM10 levels. (C) 2007 Elsevier Ltd. All rights reserved.