A headspace solid-phase microextraction (HS-SPME) method in combination with gas chromatography-mass spectrometry (GC-MS) has been used for extraction and identification of components of the volatile fraction of "Provola dei Nebrodi", a typical semi-hard Sicilian cheese. Cheese samples from different producers and at different ripening stages have been examined. The effects of various conditions (e.g. sample volume, sample headspace volume, sample heating temperature, extraction time, etc.) on extraction efficiency were studied in order to optimise the technique. The technique used made it possible to identify 61 components: fatty acids from C(2) to C(14) and their esters, aldehydes, alcohols, methyl ketones, delta-lactones, aromatic compounds, hydrocarbons and terpenes. The main components were butanoic, hexanoic and octanoic acids. The linear free fatty acids (FFA) from C(2) to C(10) were quantified by using the standard addition method. Calibration curves constructed for the FFA spiked into cheese were highly linear with a correlation coefficient R(2) > 0.998. Significant statistical differences (P> or =0.05) were evident for the even-carbon-number fatty acids during ripening.