The objective of this study was to demonstrate the effectiveness of a lightning detection network in providing a real-time data base from which automated observations of thunderstorms and lightning in the vicinity of an air base can be derived. A direction-finding cloud-to-ground lightning detection network (LDN) operated by the State Univ. of New York at Albany provided a data base of CG lightning locations in the northeastern United States for July-August, 1986. Copies of the original weather observation forms from 25 AF weather stations in the same area and time period provided a data base of thunderstorm and lightning observations for comparison. The lightning data base was analyzed for strike events located within 15 nm of each weather station. If the 15 minute period lapsed without another strike, the event was ended. Events were categorized as single- or multiple-strike events and the duration of the events tabulated. The weather observation data base was analyzed for thunderstorm events. The beginning time, ending time and duration of the events were tabulated as well as the type of lightning that was reported with each observation. Chronological comparison of the two data bases, using the weather observation data base as 'ground truth', generated 4 X 4 contingency tables of the combinations of thunderstorm-lightning events. Four parameters were derived from the statistical evaluation of the contingency tables: 1) Probability of detection; 2) False alarm rate; 3) Missed threat rate; and 4) Critical success index. Two case studies were performed with the lightning strike data base.