During the TRMM-LBA experiment a microburst occurred directly over the Profiler radar (PR). Profiler and S-POL radar observations provided a unique set of observed physical conditions to explain its origin. Neither radar showed a BB; the Z profile was consistent with hail and graupel; the linear depolarization ratio showed no melting snow. Maximum hail size of 0.8 to 0.9 cm was deduced from the PR Doppler spectra. S-Pol showed the invasion of warm precipitation-free air at the back of the storm cell setting up the sharp contrast between the cooling by melting hail and the environment. The spectral width of the PR spectra exceeded 3 m s-1 down to a height of 3 km, or two kilometers below the 0°C level. Such broad spectra can not be explained by the drop size distribution of rain alone. Only a combination of the rain from hail melt and the residual larger hail can account for the spectral width. Below 3 km the spectral width decreased to that associated with rain alone. Peak rain rates at the surface reached 124 mm h-1 and concentrations of 4.2 g kg-1 at 1724 UTC from the melt water thus providing substantial negative buoyancy. The Srivastava (1987) model shows that the dominant effect occurs near the bottom of the melting layer where condensation on the cold (0°C) particles enhances the heat transfer and melting. Negative buoyancy is due in roughly equal parts to the cooling and the weight of the melting hail. This occurs near the 3 km height and causes the air to accelerate sharply below 3 km, reaching a peak downdraft of ª15 m s-1 at the 1.5 km level at 1726 UTC. The surface wind at the Profiler turned sharply from SE to north at 1728 UTC corresponding exactly to the time of arrival of the downdraft. A moderately strong wind burst of 15 m s-1 followed at 1732 UTC. The occurrence of the microburst at the backside of the storm cell where the negative buoyancy is maximized in relation to the adjacent ambient air suggests that this is a likely region to watch for such severe events.