[1] Reflectance measurements from the spaceborne Polarization and Directionality of Earth Reflectances (POLDER) instrument are used to analyze the so-called hot spot directional signature in the backscattering direction. The hot spot is measured with an angular resolution better than half a degree using the directional capabilities of the radiometer, with some assumptions on the spatial homogeneity of the surface. The analysis yields the first quantitative observation of the hot spot signature of vegetated surfaces, with such angular resolution. The measurements show that the hot spot reflectance is a function of the phase angle x rather than a function of a parameter Á, often used in hot spot modeling, that quantifies the horizontal distance between Sun and view directions. The observed directional signature is very accurately fitted by a linear ratio of the phase angle, as predicted by a simple theory of radiative transfer within the canopy foliage. Most of the measured hot spot half widths are between 1°and 2°. Some dispersion occurs for the cases belonging to the forest and desert International Geosphere-Biosphere Program (IGBP) classes, in the range 1°to 5°. Theory predicts that the width is independent of wavelength. Our measurements indicate that the widths at 670 and 865 nm are very close, but with a significant scatter in regards to the rather small variability. The distribution of the width as a function of the IGBP surface classification shows a variability within the classes that is larger than between the classes, except for the ''evergreen broadleaf'' class. The hot spot reflectance amplitude is generally on the order of 0.10-0.20 at 865 nm and 0.03-0.18 at 670 nm, although the full range of values is wider. For thick canopies, it may be interpreted in terms of foliage element (leaf ) reflectance. Retrieved values are on the order of 0.4 in the near infrared and in the range 0.05-0.20 at 670 nm. At 440 nm, the amplitude of the signature is very small, as is expected from the small surface reflectance. This confirms that the atmospheric contribution to the reflectance increase at the backscattering direction is negligible.