The remediation of former industrial sites is, for many reasons (health and environmental impact, land use...) of great concern. Soils encountered at former coking plants are often impacted by PAHs inherited from coal-tar storage and dissemination. The present work focuses on improving the effectiveness of in situ chemical oxidation (ISCO) to reduce DNAPL (Dense Non-Aqueous Phase Liquids) remaining after pumping in the saturated zone of the impacted soil. The study is part of the BIOXYVAL project, dedicated to the remediation of contaminated sites by coordi- nating various innovative techniques on a former coking plant in the north east of France. The efficiency of seven different chemical oxidants was compared through batch experiments. PAH degradation and formation/ elimination of polar Polycyclic Aromatic Compounds (PACs) were monitored. Polar PACs including oxygen/nitrogen/sulfur- containing PACs, known to be oxidation by-products, have greater mobility in groundwater compared to PAHs and can be more toxic. Conventional oxidants including potassium permanganate, heat-activated persulfate, Fenton-like (activated with magnetite) and Fenton reactions were compared to an innovative and very promising oxidant: potassium ferrates (FeVI). Their reaction products consist of iron hydroxides (FeIII), they have also coagulant properties and are expected to be less harmful to the environment than other end-products (sulfuric acid, manganese salts). Addition- ally, ferrates were used as a Fenton-like catalyst under acidic or neutral conditions. Oxidations were performed on a reference matrix (Fontainebleau sand) spiked with DNAPL sampled at the studied site, at three reaction times (3 h, 1 day and 1 week). Oxidants were applied at one stoichiometric oxidant demand (SOD) deduced from the elemental composition of the DNAPL. Magnetite Fenton-like oxidation applied on spiked sand was not very effective (28.5% degradation of 16-US EPA PAH). The ferrates managed to reach 77.9% degradation of PAH after one week even if the protocol was not optimized. The best oxidants were KMnO4 and heat-activated persulfate, which decreased up to 98% PAH and 96.1% PAH, respectively. Similar results were obtained on the extractable organic matter, representing the whole DNAPL pollution. Good degradation rates were also obtained with classical Fenton reaction and Ferrates/Fenton at pH 3: 93.9% and 89.8% PAH removal respectively. Some O-PACs were formed as degradation by-products in different proportions depending on the oxidants and conditions. Some oxidants are currently optimized in laboratory column tests in order to select the one that will, eventually, be applied at pilot scale. The first step consists in spiking glass beads or soil with DNAPL by drainage/imbibition, followed by water flushing to mimic the removal of the DNAPL fraction by pumping. This will give a residual saturation (Srn) of tar similar to the one obtained on field. Then Ferrates or KMnO4 are flushed through the column. The water quality is monitored throughout the experiment and the PACs remaining in the matrix after treatments are quantified.