Military installations, airports, sporting events, and other facilities curtail operations when cloud-to-ground (CG) lightning is present. The National Lightning Detection Network records approximately 20 million lightning flashes each year (Orville and Huffines 1999). Because of the frequency and random nature of CG lightning more people become casualties to lightning each year than to either tornadoes or hurricanes. Lightning specific warning criteria are not standard and appear to have evolved over time as a result of increasing the distance in response to lightning incidents until the proper balance between threat and impact were achieved, rather than through research and lightning data analysis. This research effort attempted to quantify what constitutes a safe distance when lightning is present. The method used in this research project groups lightning flashes into clusters using spatial and temporal constraints. However, not all flashes meet the time and distance criteria for clustering and remained outside of the grouped flashes and as such are identified as isolated flashes. These isolated flashes are outliers in the data set, but are precisely the flashes that prove most dangerous. For this reason not only were the distances between each flash and cluster center studied, but also the distances between each isolated flash and its nearest neighboring flash. Distributions for both distances were studied for the continental U.S. by season.