Polycyclic aromatic hydrocarbons (PAHs) are well-known environmental contaminants and 15 of which exert both genotoxic and carcinogenic properties. These compounds can be found in food at residue levels, result of a contaminated environment or the type of food processing applied such as smoking or roasting. In addition, Butler et al. has demonstrated that food is the major exposition route of the general population to benzo[a]pyrene (B[a]P). To protect human health, many competent Authorities has established maximum residue levels in food. In Europe, a MRL is fixed for B[a]P, considered as a marker for PAH contamination, and in 2012 the legislation will concern 3 more PAH: benzo[a]anthracene B(a)A, benzo[b]fluoranthene (B[b]F) and Chrysene (Chr). Therefore, food samples will have to be analysed to demonstrate compliance with these limits. The analysis of PAH has been mostly focussed on environmental matrices and little is published about food. In 2008, Windal et al. reported a method for the analysis of the 16 European polycyclic aromatic hydrocarbons (PAHs) in vegetable oils and plant extract using a donor acceptor complex chromatography (DACC). This DACC clean-up is highly specific to PAHs and was then, tested for different type of food matrices. The applicability of this technique to all food matrices was hampered by some co-extracts not separated by the column and interfering with the fluorescence detection of the PAHs. This research deals with the evaluation of different type of clean-up addition to the DACC to enable easy determination of the 16 PAHs. All the tests were performed with fish oil because this matrix, one of the most difficult food matrixes, displayed after DACC clean-up many interferences. First, we compared the GPC (Gel Permeation Chromatography) on a polymeric phase to the DACC clean-up. In terms of the solvent consumption and clean-up efficiency, DACC demonstrated better performances and was therefore kept in the protocol. Then, different types of SPE were tested as a post DACC clean-up step. Phenyl and C18 cartridges were used as chemical filters: the DACC Acetonitrile eluate was directly loaded. In the opposite, normal phase such as silica and SupeIMIP® required a solvent exchange prior to loading. We will show that chemical filters are not efficient for the removal of interferences. For the normal phase SPE cartridges, the SupeIMIP® gave better results than silica. All SPE options could not eliminate early eluting interferences which hamper the determination of benzo[c]fluorene (B[c]F). However, in the elution range of the 4 PAHs under legislation in the next future, several co-eluting interferences have been efficiently removed.