The Terminal Doppler Weather Radar (TDWR) has an operational requirement to provide a one minute advance warning for aircraft encountering a hazardous wind shear. This paper describes the use of features aloft in the prototype TDWR microburst recognition algorithm to improve the timeliness of microburst hazard warnings. The use of features aloft allows the algorithm to make a microburst declaration while the surface outflow is still weak, thereby increasing the hazard warning time. In addition, current work indicates that these signatures can also be used to predict the onset of surface outflow for high-reflectivity events. An initial version of the microburst recognition algorithm using surface velocity data only was described by Merritt (1987). Initial work on the use of features aloft to increase the reliability and timeliness of microburst alarms was described in Campbell, 1988. This work was motivated by the desire to emulate the ability of human experts to use features aloft to enhance the timeliness of microburst warnings (McCarthy & Wilson, 1986). This research was further influences by the conceptual models for the evolution of low, medium and high reflectivity microburst events in the Denver area proposed by Roberts and Wilson (1986), and by studies of features aloft associated with microbursts in the Southeast (Isaminger, 1987). The current TDWR microburst recognition algorithm is described in Campbell and Merritt, 1988. The present paper presents results demonstrating the ability of the prototype algorithm to recognize features aloft for microburst events observed at Huntsville, AL and Denver, CO. It is shown that the ability to recognize features aloft improved the hazard warning time for these events. Initial results for microburst prediction are also presented.