A 3 year set of PM10 monitoring data using the tapered element oscillating micro-balance from 3 sites in London was analysed, in conjunction with air stream back trajectories and meteorological data for the same 24 h periods. Geometric mean PM10 for any site against groupings of trajectory wind direction (TWD) showed a south-easterly peak in PM10, Analysis of variance was carried out using London meteorological parameters and also wind speed and direction information calculated from positions on air mass back trajectories at 4 different periods prior to the approach to London. This model explained 60-65% of the observed variation, and because the analysis enables the meteorological factors to be de-coupled from the TWD, it can be concluded that local weather is not the prime cause of the peak. It was found that for 6-7% of the time when the TWD was from 120 degrees -160 degrees at least 15-20 mug m(-3) of extra PM10 was advected in. Evidence is presented that the local meteorology is responsible for re-suspension. By removing individually each parameter retained in the model, its relative contribution to explaining variance was assessed. TWD, then trajectory wind speed (TWS) were found to be the most important. There was a calculated reduction of 7.0 mug m(-3) (95% confidence 6.0-8.0) of PM10 over the 1047 day monitoring period. (C) 2001 Elsevier Science Ltd. All rights reserved.