Two procedures have been investigated for the quantification of the different forms of mercury in food. A two-stage procedure has been developed to determine firstly total inorganic and organometallic species, and then the full separation of all organomercury species. The procedure involves solubilisation of the samples using alkaline extractions or enzymolysis, followed by the extraction of organic mercury in an organic solvent, preferably a mixture of dichloromethane and hexane (3:2). For the total organic mercury determination, the organic extract is analysed for "total" mercury after nitric acid/peroxide digestion, evaporation of the solvent and detection by cold vapour-atomic fluorescence spectrometry. Full organomercury speciation requires a clean-up step before analysis of the final extract in dichloromethane by gas chromatography coupled to a pyrolyser and an atomic fluorescence detector (GC-pyro-AFS). A detection limit of 6 ng l-1, and reproducibility of 2% was achieved for the CV-AFS method; GC-pyro-AFS yielded 200 ng l-1 and 5% for detection limit and coefficient of variation, respectively. Both procedures were validated with the use of various certified reference materials over a wide range of mercury concentrations, and by spiking experiments. The validated methods were tested successfully on a wide range of commercially available food samples.