Five measurement campaigns were conducted in southern Arizona (AZ), northern Texas and southern Oklahoma (TX-OK), and in the Great Plains (GP) of eastern Colorado, western Kansas and western Nebraska to record lightning using video cameras that were synchronized to GPS time. By comparing the video recordings with reports from the U.S. National Lightning Detection NetworkTM (NLDN), we have documented how the characteristics of confirmed cloud-to-ground (CG) flashes and strokes vary with geographic region. The results show that the estimated peak current (Ip) of negative first strokes, as measured by the NLDN, varied considerably between different recording sessions and that the distributions of Ip and the multiplicity of negative flashes also varied with region. Approximately two-thirds of the CG flashes in the GP were positive, while in TX-OK and AZ less than 10% of CG flashes were positive. For the GP, radar imagery was combined with NLDN reports to show when and where in the storm development the positive and negative flashes occurred, and to determine if the flashes we recorded were biased by the sampling. In this presentation, we will summarize the radar results in the GP and the parameters of positive and negative CG flashes in all three regions. A key result is that the median values of Ip for positive and negative first strokes in all three regions are significantly different from values commonly found in the engineering literature.