Wind shear is a major cause of aircarrier accidents in the United States, and most of these accidents have been caused by one particular form of wind shear called a microburst (Zorpette, 1986). Microbursts have been defined as small scale, low-altitude, intense downdrafts which impact the surface and cause strong divergent outflows of wind. We know they are associated with thunderstorms and are usually but not always accompanied by heavy rainfall at the ground. However, a number of meteorologically distinct phenomena associated with thunderstorms can give rise to strong downdrafts and high surface winds. Most microburst research has focused on the main precipitation driven downdraft of thunderstorms, both with and without significant surface rainfall. But other downdraft types such as the dynamically driven downdrafts at low altitude associated with "vortices" at the leading edge of expanding thunderstorm outflows and with "roll clouds" have also been associated with the microburst problem. In this paper, I discuss these two primary forms of low altitude downdraft phenomena in thunderstorms. This differentiation is essential to discovering exactly what atmospheric conditions lead to the development of the most hazardous microbursts. A physically based predictive model for thunderstorm downdraft strength is presented which shows that the radar reflectivity of a storm alone cannot be used as a hazard index; information about the static stability of the atmosphere is also essential. I then show that the downdrafts associated with the gust front around a cold outflow from a small isolated thunderstorm, a microburst, are inherently stronger at low altitudes than those found in more straight-line gust fronts. Finally, I reexamine the most recent fatal U.S. microburst accident, the crash of Delta 191 at Dallas/Ft. Worth in 1985, and show that both types of low altitude downdrafts were encountered as part of the "microburst", although the downdrafts came from different storms.