Data mining tools were used to examine the use of atmospheric stability indices as a means to forecast lightning occurrence and the degree of lightning expected at a station for the next 12 hours. The most useful radii of CG lightning summaries around a station was determined to be 50nm and these data were then examined against all upper-air station data throughout the Midwest. Traditional statistical methods were used to establish the most related stability indices, which turned out to be the CAPE, K, KO, Lifted, Showalter, Total Totals, and SWEAT indices. Cloud-to-ground (CG) lightning summaries from the National Lightning Detection Network (NLDN) from 1993 thru 2000 were studied to determine the indices usefulness as predictive tools for determining CG lightning activity. Also explored was the improvement upon the commonly accepted thresholds of the stability indices as general thunderstorm indicators. An improvement was found and new threshold ranges were developed for relating stability index values to lightning occurrence (thunderstorms). Since traditional regression methods failed to find significant predictive relationships between the indices and CG lightning occurrence, the detection and classification abilities of decision trees derived from classification and regression tree analysis (CART) best served the purposes of this study. Decision trees were examined on the large available databases and significant results were found, resulting in the development of a lightning forecast tool for both the probability of lightning occurrence and the intensity of the occurrences once forecasters are convinced of a high enough probability. The predictive ability of the decision trees used in this study for lightning detection often exceeded 80-90% for most locations with a high degree of significance. The decision tree results were formulated into a forecast prediction tool with summary results for each location. These results are specified both graphically and textually in a user-friendly format for forecasters to use as a “ready to use” predictive tool for forecasting lightning activity. Forecasters in the Midwest U.S. are encouraged to use this “innovative” forecast tool immediately for forecasting lightning (and thunderstorm) activity.