A procedure was developed to detect hazardous regions within thunderstorms using weather radar data. The procedure is based upon the hypothesis that convective turbulence occurs within 2-3 km of relative radar reflectivity maxima (cells). The hypothesis was tested using a limited set of simultaneous aircraft and radar data. Good agreement was found between the observed regions of convective turbulence and the cell locations determined by radar. The viability of the hazard detection hypothesis as a basis for automatic warning and forcast depends upon the reliability of the cell detection and tracking algorithms. Analysis of precision radar data revealed that the cells are small in area extent, have a detection probability in excess of 0.9 using multiple radar scans, and are readily tracked for periods between 10 and 20 minutes. The characteristics of radar systems for acquiring data to support cell detection, are discussed. The role of Doppler spectral data is explored, and it is found that practical limitations on radar beamwidth hamper direct observation of turbulence on the scale size hazardous to aircraft.