Abstract: Hailstones are products of strong convective cells, characterized by locality and shortlived feature and hence studies of hail clouds are more or less constrained if conventional surface and upperlevel observations are employed. In recent years, with advances in radar technology, the detection items and space/time resolutions have increased, the digitalized radar has now become one of the main tools in exploring hail cloud development. As early as the 1970s, a prognostic factor i.e. vertically integrated liquid water content (VIL) was proposed for studying digitalized radar echo data, and it was defined in calculation as the vertical integration of mixing ratios of the liquid water content, which was found from the empirical relation between radar measured reflectivity factor and raindrop concentration. It is well known that radar VPPI data are discontinuous from one level to another; the number of levels for elevation angles is too small, significant noises are produced in radar operation, and furthermore the reflectivity factor of the cell is nonlinear so that direct vertical integrationwould bring about greater errors. For this reason, based on the constant level data such as the reflectivity factor of new generation weather radar obtained with the 3DBarnes interpolation scheme, the VIL during the evolutional processes of hail cells is calculated using its theoretical model, and the maximum of VIL(VILmax) is identified by use of MAX function. With statistic and piecewise function techniques, the evolutional character of VILmax and its relation with hail shooting time on the ground are detailedly analyzed for 16 hail cells in the northeast Tibetan plateau, during May-August of 2004-2005. Results show that there were apparently spatialtemporal differences in VILmax among different hail cells at the hail shooting time, but similar “explosive increasing" and “explosive decreasing" phenomena around the time in the evolution process of each hail cell; two “explosive increasing" processes appeared within the four time intervals ofradar data (22 min) before the first hail shooting, with no shooting on the ground after the first VILmax increasing until the occurrence of second increasing about 1-2 time intervals (5-11 min)after the first increasing; there was no first “explosive increasing" before the second shooting in the same cell, and the hail shooting on the ground stopped with the occurrence of “explosive decreasing"; the positive(negative) peak of GVILmax (which is defined as the change rate of VILmax, briefed as GVILmax) corresponded completely with the “explosive increasing (decreasing)" in time, and therefore is a good indicator for hail shooting; the time error of the empirical relation between the peaks and the time of hail shooting on the ground is within one time interval (5-6 min).