PM10 and PM2.5 chemical composition has been determined at a suburban insular site in the Balearic Islands (Spain) during almost one and a half year. As a result, 200 samples with more than 50 chemical parameters analyzed have been obtained. The whole database has been analyzed by two receptor modelling techniques (Principal Component Analysis and Positive Matrix Factorisation) in order to identify the main PM sources. After that, regression analyses with respect to the PM mass concentrations were conducted to quantify the daily contributions of each source. Four common sources were identified by both receptor models: secondary nitrate coupled with vehicular emissions, secondary sulphate influenced by fuel-oil combustion, aged marine aerosols and mineral dust. In addition, PCA isolated harbour emissions and a mixed anthropogenic factor containing industrial emissions; whereas PMF isolated an additional mineral factor interpreted as road dust + harbour emissions, and a vehicular abrasion products factor. The use of both methodologies appeared complementary. Nevertheless, PMF sources by themselves were better differentiated. Besides these receptor models, a specific methodology to quantify African dust was also applied. The combination of these three source apportionment tools allowed the identification of 8 sources, being 4 of them mineral (African, regional, urban and harbour dusts). As a summary, 29% of PM10 was attributed to natural sources (African dust, regional dust and sea spray), whereas the proportion diminished to 11% in PM2.5. Furthermore, the secondary sulphate source, which accounted for about 22 and 32% of PM10 and PM2.5, is strongly linked to the aged polluted air masses residing over the western Mediterranean in the warm period. (C) 2013 Elsevier B.V. All rights reserved.