We employ L-band airborne synthetic aperture radar (SAR) polarimetry to detect surface changes produced by the Tsaoling landslide, the largest slope failure triggered by the September 1999 Chi-Chi earthquake in central Taiwan. Imaging polarimeters provide a complete description of the scattering properties of radar target materials. Resurfacing of forested hillslopes by landslides alters scattering mechanisms from those dominated by backscatter from trees to mechanisms associated with scatter from rough, bare surfaces. Scattering mechanism information on a per-pixel basis is extracted by decomposing matrices formed as the outer product of the complex scattering vector measured for each resolution cell. We classify surface cover type and thereby identify the extent of the landslide, using such polarimetric parameters as scattering entropy, anisotropy, and pedestal height, derived from the eigenvalues of the decomposition, as well as the weighted average scattering mechanism derived from the eigenvectors. We address the utility of full polarimetry versus dual polarimetry for landslide mapping purposes and show that fully polarimetric SAR is necessary for distinguishing water surfaces of varying roughness from the bare surfaces created by landsliding. We show that scattering entropy and average scattering mechanism, for example, can be used to identify the Tsaoling landslide source, run out area and impounded lakes as proficiently as maps obtained using satellite optical sensors, such as Landsat and the Indian Research Satellite. However, the operational advantages of radar over optical sensing techniques (namely, its day-night, all-weather data acquisition capability) suggest that SAR polarimetry could play a leading role in the rapid assessment of landslide disasters.