Chemical composition data (ionic fraction, carbonaceous compounds and Polycyclic Aromatic Hydrocarbons) for PM2.5 were acquired during 2005–2007years by seasonal sampling campaigns performed at six urban sites in Apulia Region. In addition, an industrial site was considered. Discriminant Function Analysis, Principal Component Analysis and Absolute Principal Component Scores were applied to dataset in order to identify PM2.5 sources and to estimate their contributions to samples mass. Three main sources were obtained for the urban sites: combustion processes, natural sources and secondary particulate. Moreover, the application of Principal Component Analysis on chemical and meteorological data highlighted the influence of temperature and relative humidity on natural sources, and the relevance of pressure on secondary particulate content. In PM2.5, secondary particulate contributed on average with 78% to total mass, while combustion processes and natural sources accounted, respectively, 15% and 6.5% of the fine fraction mass. For this reason the secondary background should be taken into account in order to suggest the efficient abatement strategies to improve local air quality in South of Italy.