In this study, 437 days of 6-daily, 24-h samples of PM 2.5, PM 2.5–10 and PM 10 were collected over a 12-month period during 2003–2004 in Melbourne, Sydney, Brisbane and Adelaide. The elemental, ionic and polycyclic aromatic hydrocarbon composition of the particles were determined. Source apportionment was carried out by using the positive matrix factorisation software (PMF2). Eight factors were identified for the fine particle samples including ‘motor vehicles’, ‘industry’, ‘other combustion sources’, ‘ammonium sulphates’, ‘nitrates’, ‘marine aerosols’, ‘chloride depleted marine aerosols’ and ‘crustal/soil dust’. On average combustion sources, secondary nitrates/sulphates and natural origin dust contributed about 46%, 25% and 26% of the mass of the fine particle samples, respectively. ‘Crustal/soil dust’, ‘marine aerosols’, ‘nitrates’ and ‘road side dust’ were the four factors identified for the coarse particle samples. On average natural origin dust contributed about 76% of the mass of the coarse particle samples. The contributions of the sources to the sample mass basically reflect the emission source characteristics of the sites. Secondary sulphates and nitrates were found to spread out evenly within each city. The average contribution of secondary nitrates to fine particles was found to be rather uniform in different seasons, rather than higher in winter as found in other studies. This could be due to the low humidity conditions in winter in most of the Australian cities which made the partitioning of the particle phase less favourable in the NH 4NO 3 equilibrium system. A linear relationship was found between the average contribution of marine aerosols and the distance of the site from the bay side. Wind erosion was found associated with higher contribution of crustal dust on average and episodes of elevated concentration of coarse particles in spring and summer.