In laboratory conditions, reconstructed or intact litter samples can be used to assess the flammability of dead surface fuels but the sampling method could affect litter flammability results. To assess this effect, samples of litters were collected in south-eastern France, according to two different methods previously used in other studies, one keeping intact the structure of litter layers (intact litter) and the other using mainly the surface litter layer to reconstruct the litter sample (reconstructed litter). The sampling occurred in representative ecosystems of the study area: pure Pinus halepensis stands, mixed pine-oak stands, pure cork oak stands and in two types of shrubland (maquis and garrigue). Flammability experiments were carried out on fire bench, using a glowing firebrand as ignition source and a 10 kmh-1 wind. The main variables recorded were: ignition frequency, time-to-ignition, flaming duration, flame rate of spread, flame propagation, mean flame temperature, maximum flame height and rate of consumption. The comparison of flammability results showed that the sampling method had a significant effect on litter bulk-density, rate of spread and rate of consumption, intact litter being more flammable than reconstructed litter. The type of vegetation had a significant effect on litter depth, ignitability and combustibility (except on flame temperature and rate of spread) and the litter composition could explain in part this fire behaviour. The effect of the reconstruction of litters on flammability parameters and its magnitude differed according to vegetation types. The intact structure of the litter appeared to be an important driver of its flammability, especially for combustibility and consumability. The assessment of these parameters will differ when using reconstructed litter samples compared to intact litter samples. Future research on litter flammability should take into account the bias due to the litter sampling method when the litter is reconstructed.