Epidemiological studies have found negative associations between human health and particulate matter in urban air. In most studies outdoor monitoring of urban background has been used to assess exposure. In a field study, personal exposure as well as bedroom, front door and background concentrations of PM(2.5), black smoke (BS), and nitrogen dioxide (NO(2)) were measured during 2-day periods in 30 subjects (20-33 years old) living and studying in central parts of Copenhagen. The measurements were repeated in the four seasons. Information on indoor exposure sources such as environmental tobacco smoke (ETS) and burning of candles was collected by questionnaires. The personal exposure, the bedroom concentration and the front door concentration was set as outcome variable in separate models and analysed by mixed effect model regression methodology, regarding subject levels as a random factor. Seasons were defined as a dichotomised grouping of outdoor temperature (above and below 8 degrees C). For NO(2) there was a significant association between personal exposure and both the bedroom, the front door and the background concentrations, whereas for PM(2.5) and BS only the bedroom and the front door concentrations, and not the background concentration, were significantly associated to the personal exposure. The bedroom concentration was the strongest predictor of all three pollution measurements. The association between the bedroom and front door concentrations was significant for all three measurements, and the association between the front door and the background concentrations was significant for PM(2.5) and NO(2), but not for BS, indicating greater spatial variation for BS than for PM(2.5) and NO(2). For NO(2), the relationship between the personal exposure and the front door concentration was dependent upon the "season", with a stronger association in the warm season compared with the cold season, and for PM(2.5) and BS the same tendency was seen. Time exposed to burning of candles was a significant predictor of personal PM(2.5), BS and NO(2) exposure, and time exposed to ETS only associated with personal PM(2.5) exposure. These findings imply that the personal exposure to PM(2.5), BS and NO(2) depends on many factors besides the outdoor levels, and that information on, for example, time of season or outdoor temperature and residence exposure, could improve the accuracy of the personal exposure estimation.