This paper describes a prototype microburst prediction product for the Terminal Doppler Weather Radar (TDWR). The prediction product was evaluated for microbursts observed during the spring and summer of 1989 at Kansas City. Results are presented demonstrating reliable prediction of high reflectivity microbursts of at least 15 m/s outflow intensity from single-Doppler radar data. The ability of the algorithm to predict microbursts approximately five minutes prior to the onset of surface outflow could be used to improve air traffic control (ATC) planning and to improve hazard warning time to pilots. In particular, this product could allow aircraft to avoid an impending microburst hazard, rather than penetrating it. The present TDWR microburst recognition algorithm uses features aloft such as reflectivity cores and convergence to recognize microburst precursors. The algorithm uses precursors to make a microburst declaration while the surface outflow is still weak, thereby improving the hazard warning time (Campbell, 1989). The microburst prediction product is an extension of the algorithm to predict microbursts from these precursor signatures. The prototype prediction product is tuned to predict the high reflectivity microburst typical of humid regions of the United States. The paper begins by reviewing conceptual models for microburst development and comparing them to the observed characteristics of Kansas City microbursts. The prototype prediction product is then described, and performance statistics are presented. Finally, failure mechanisms and future work are discussed.