A new generation of models of a forest fire is currently being developed; they include more and more physical mechanisms. The main objective of the present study was to provide experimental measurements to test such models. An apparatus was designed to simulate and study in laboratory conditions the flame and the near-field plume stemming from the combustion of an isolated shrub. The burner was made of a cylindrical wire mesh basket filled with a forest fuel ignited at the lower circumference of the basket. Three diameters (20, 28, and 40 cm) for the basket and two kinds of fuel (Pinus pinaster needles and excelsior) were used. Temperatures were measured inside and above the burner using type K thermocouples of 50-mum diameter. Three pairs of these sensors were especially used to estimate the upward gas velocity, thanks to the expected cross-correlation of thermal fluctuations. We obtained varied and non-steady regimes of combustion as desired, in particular the maximum heat release rates ranged between 30 and 180 kW. The structure of the visible flame and its temperatures were analyzed for a fully developed flamed The height of a flame was found to scale with the heat release rate according to the usual two-fifths power law, which enables a normalized flame height to be defined. Vertical temperature profiles in the flame were found to scale with the normalized height and radial temperature profiles fitted Gaussian laws well. The determination of gas velocity was uncertain, but the usual scaling also applied to the measurements. Scaling rules, which apply well to either steady turbulent diffusion flames on a porous gas burner or pool fires, hold for the non-steady flame on our forest fuel burner, when observing a fully developed flame. (C) 2003 The Combustion Institute. All rights reserved.