Principal component analysis (PCA) coupled with a multilinear regression analysis (MLRA) was applied to PM(10) speciation data series (2002-2005) from four sampling sites in a highly industrialised area (ceramic production) in the process of implementing emission abatement technology. Five common factors with similar chemical profiles were identified at all the sites: mineral, regional background (influenced by the industrial estate located on the coast: an oil refinery and a power plant), sea spray, industrial 1 (manufacture and use of glaze components, including frit fusion) and road traffic. The contribution of the regional background differs slightly from site to site. The mineral factor, attributed to the sum of several sources (mainly the ceramic industry, but also with minor contributions from soil resuspension and African dust outbreaks) contributes between 9 and 11 microg m(-3) at all the sites. Source industrial 1 entails an increase in PM(10) levels between 4 and 5 microg m(-3) at the urban sites and 2 microg m(-3) at the suburban background site. However, after 2004, this source contributed less than 2 microg m(-3) at most sites, whereas the remaining sources did not show an upward or downward trend along the study period. This gradual decrease in the contribution of source industrial 1 coincides with the implementation of PM abatement technology in the frit fusion kilns of the area. This relationship enables us to assess the efficiency of the implementation of environmental technologies in terms of their impact on air quality.