Four commercial quartz dusts (flours), two inflammogenic in vivo and activating macrophages in vitro (Qz 2/1-c and Qz 3/1-c) and two mostly inert (Qz 5/1-c and Qz 11/1-c), have been compared regarding their surface properties, in order to detect chemical differences which may account for their different biological behaviour. The following features have been examined: 1) extent of the amorphous fraction (heat associated alpha<-->beta transition of quartz) and its solubility in HF; 2) potential to cleave a carbon-hydrogen bond with consequent generation of carbon centred radicals (spin trapping technique, EPR); 3) evolution of surface functionalities upon heating (FTIR spectroscopy); 4) mechanisms of adsorption of water on dusts outgassed at 150 degrees and at 800 degrees C (adsorption calorimetry). HCl treated samples have also been examined. The two "less toxic" quartzes are more resistant to HF attack, coordinate irreversibly H2O molecules and exhibit strong adsorption sites, which are absent in the other two and in a very pure quartz dust. Conversely all samples show the same potential to release free radicals. The different behaviour of the two sets of dust is consistent with a different level of impurities, namely aluminium ex kaolin, carbon and alkaline ions. The less inflammogenic quartzes appear to be covered by aluminium ions (and possibly iron) which strongly holds molecular water or carbonates, thus reducing the silanol patches to a large extent and changing the surface properties of the particles. We hypothesize that cellular response, and particularly macrophage activation and death, is mediated by strong interactions between silanol patches and some cell membrane components, but inhibited when the surface of the particle is modified by the presence of aluminium ions, surface carbonates and other metal contaminants. This hypothesis suggests that grinding procedures with little appropriate additives, e.g. kaolin, alumina, can reduce the biological activity of quartz dusts.