Thermal analysis have widely been used for the characterization of various materials, but only few studies concern its application to atmospheric particulate matter (Matuschek et al., 2004; Duarte and Duarte, 2008). All these studies are focused on the identification or quantification of desorbed organic species. Among inorganics, the species that are expected to thermally desorb from PM are ammonium salts and water. Ammonium salts are subjected to solid/vapour equilibria that are responsible for one of the main artefacts in PM determination. The extent of these artefacts has been proved to be influenced by water vapour, which may be absorbed on particles. In the scientific literature, however, only a few and contrasting results have been reported about the analytical determination of water in PM samples, and they have been mostly obtained by the Karl Fischer method (Otha et al., 1998; Tsai and Kuo, 2005). In this work, the thermal behaviour of water and ammonium salts has been examined under controlled conditions, with the purpose of exploring the potential of thermogravimetry and thermal desorbtion techniques with respect to both the analytical determination of water and the study of sampling artifacts mechanisms. The themogravimetric curves of reference materials NIST 1648 and NIST 1649a (figure 1) show, as expected, different steps of mass loss. In the range from 25°C to 180°C a loss of ca. 6% (6.3% for NIST 1648 and 4.9 % for NIST 1649a) in two steps has been evidenced. These losses have been assigned, respectively, to weakly and strongly bounded water by using hyphenated TG-MS and TG-FTIR and by carrying out the chemical analysis of the residual obtained after heating at different temperatures. Although the method needs further validation, the thermo-gravimetric determination of PM water content seems to be a convenient alternative to the use of Karl Fischer method.Despite the evidence of ammonium salts evolution during the sampling at ambient temperature, the loss of ammonium salts from reference materials starts above 180°C and the residual concentrations of nitrate, chloride and ammonium become negligible at about 310°C. Mass loss registered by the TG analysis in the 180-310 °C range is 7.2% for NIST 1648 and 7.9 % for NIST 1649a. This loss is partially due to inorganic salts (i.e. for NIST 1648: ammonium 1.8%, nitrate ca. 1,1%, chloride 0.3% and sulfate, that partially decomposes to SO2, ca. 1.1%) and partially to organic matter. The latter has been estimated to be about 2.5% by EC/OC thermoanalysis carried out in the same temperature range. Similar results have been obtained also on real PM10 samples collected on Teflon membranes at an urban site in Rome, Italy. Even though the presence of the filter causes a worsening of signal to background ratio, also in this case water loss at T