PM2.5 speciation for the major chemical components has been performed on 109 daily samples collected by a high-volume gravimetric automatic sequential sampler. The 24-h sampling campaigns have been performed at an urban background (UB) site and at a tunnel site in the city of Milan. PM2.5-loaded filters are analysed for the main ionic components (chloride, nitrate, sulphate, ammonium) and the carbon species, elemental carbon (EC) and organic carbon (OC). The species analysed identify on average about 90% of the total mass sampled in the UB site and about 80% of the total mass sampled in the tunnel site. For the latter site a more complete characterisation, accounting for elemental composition, is performed: mass closure that also includes crustal fraction and metal oxides results in a slight overestimation (5%) of the total PM2.5 mass. Based on elemental composition data, the role of the potential sources of crustal and anthropogenic components is identified by the enrichment factor technique. The values obtained for the speciation are typical of urban areas and point out a strong contribution of the secondary source and a small contribution from the emissions of stationary sources in winter. A 0.9 value for the ratio between organic compounds and elemental carbon in PM2.5 samples collected at the tunnel site is observed. Applying the OC/EC ratio approach, this value, representative of the real traffic emissions for the circulating vehicle fleet in Milan, is used for the determination of primary organic aerosols due to traffic in ambient PM2.5. The primary contribution of the traffic source, in terms of carbonaceous matter (EC and primary organic aerosols), is estimated in 6% and 11% of the total PM2.5 mass, respectively, in the cold and warm season. The same OC/EC ratio approach also allows for the evaluation of secondary organic aerosols (SOA) presence in warm season PM2.5 samples: SOA is about 85% of total particulate organic matter and 30% of the total PM2.5 mass. Results obtained for the primary contribution of traffic and for SOA contribution to PM2.5 concentration levels confirm that traffic restriction interventions are not effective to provide relevant short-term air quality improvement. The need to control the emission of the precursors of PM2.5 secondary components is pointed out.