In the period 1946-2001, there were 225 armed conflicts and 34 of them were active in all of or part of 2001. Armed conflict remains a serious problem in the post-Cold War period. For three decades, the Correlates of War project has served as the main supplier of reliable data used in longitudinal studies of external and internal armed conflict. The COW datasets on war use the relatively high threshold of 1,000 battle-deaths. The Uppsala dataset on armed conflict has a lower threshold, 25 annual battle-deaths, but has so far been available for only the post-Cold War period. This dataset has now been backdated to the end of World War II. This article presents a report on armed conflict based on this backdate as well as another annual update. It presents the procedures for the backdating, as well as trends over time and breakdowns for the type of conflict. It assesses the criteria for measuring armed conflict and discusses some directions for future data collection in this area.