Thunderstorms are a significant factor in the planning and execution of Defense (DoD) operations in Afghanistan, especially in the spring and summer. Skillful forecasting of Afghanistan thunderstorms has proven difficult, even at relatively short lead times of 24 hours or less. This has led to adverse effects on a wide range of DoD missions. One potential reason for the forecasting difficulties is a lack of understanding of the conditions that lead to static instability and thunderstorms in the elevated desert mountain environment that characterizes much of Afghanistan. Much of the thunderstorm forecasting for Afghanistan is based on forecasting methods developed for the contiguous U.S. (CONUS)--for example, the use of CONUS-based static stability indices as indicators of the potential for thunderstorm development. We have investigated methods for improving thunderstorm forecasting in and near Kabul, Afghanistan, by: (1) analyzing interannual to hourly variations in thunderstorm activity; and (2) analyzing the large-scale conditions that are favorable and unfavorable for thunderstorms. We used in situ surface and radiosonde data to characterize the local conditions associated with thunderstorm variations. Our focus was on March-May, the period with the most thunderstorm activity in Kabul. We also used global reanalysis data to analyze the large-scale conditions that are favorable and unfavorable for thunderstorm development. We developed and tested two new static stability indices for use in Kabul. We also developed a large-scale circulation index to describe the regional factors that contribute to thunderstorm variations. Finally, we identified outgoing longwave radiation anomalies that occurred in specific tropical ocean basins as potential precursors for predicting thunderstorm and nonthunderstorm events at lead times of 5-15 days.