Indoor and outdoor particle size distributions, indoor-outdoor pressure difference, indoor air-exchange rate, and meteorological conditions were measured at an uninhabited apartment located in a busy street in Copenhagen during 1-month long fall, winter and spring campaigns. Particle penetration was estimated from concentration rebound measurements following HEPA filtering of the indoor air by fitting a simple deterministic model. The model included measured air exchange rates and published surface deposition loss rates. This model was then used to predict indoor particle concentration. The model predicted well the indoor concentration of coarse (1.2-4 mum) particles of outdoor origin for the fall and spring campaign. The model performed less well for the fine (0.5-1.2 mum) particle concentration and the winter campaign. The association between the ratio measured/predicted indoor concentration and factors not included in the deterministic model was analysed statistically and the result was used to determine a correction factor to the model prediction. The correction factor was found to depend on wind velocity, outdoor relative humidity, and air exchange rate. Including the correction factor reduced the ratio of the 95 percentile to the 5 percentile by an average of 26% for the fine particles and 12% for the coarse particles. The ratio measured/predicted concentration using the correction factor was found to be the highest during periods where it was most likely that occupants were present in other apartments. The results suggest that factors such as particle chemical composition, within building transport patterns, and occupant behaviour in other apartments should be identified and quantified in future studies, and that these factors need to be included in predictive models. (C) 2004 Elsevier Ltd. All rights reserved.