The aim of the present work was to improve the study of wood self-extinguishment under cone calorimeter thanks to a dedicated experimental setup. A cone calorimeter was modified into a double sliding cone calorimeter in vertical orientation for studying samples extinguishment. With this new setup, the external heat flux can be quickly switched from a high to a low value seeking for the heat flux leading to the sample extinguishment. Samples were first submitted to different heat fluxes (60, 82.5 and 93.5 ) leading to auto-ignition. After different exposure times (6, 9, 15 and 18 min), the heat flux was suddenly reduced to a lower value (between 38.5 and 60 ) by sliding from one cone to the other one. All in all, 206 tests were performed varying heat fluxes and exposure time. This large number of test made it possible to carry out a statistic study and to deduce a critical Mass Loss rate and a critical heat flux for self-extinguishment. The critical heat flux for self-extinguishment varied from 44 to 51 depending on the heat flux that was used to ignite the sample. These critical heat fluxes were found weakly dependent on the time during which the sample was exposed to the high flux. Time-to-extinguishment was between 40 s and 100 s. Mass Loss Rate Per Unit Area (MLRPUA) for which self-extinguishment occur was found to be equal to 3.95 and this value is almost independent on experiment parameters (exposure time and heat flux).