The rationale is straightforward and persuasive: intrastate conflicts are by definition subnational phenomena. If we want to understand them fully, it may be wise to refocus our attention from the country level to the subnational level. Where violence is located might inform us as to why it erupts and how it is linked to various political, economic or social factors. The number of statistical geospatial analyses undertaken at the subnational level has been increasing constantly in recent years. Even though such studies have contributed greatly to peace and conflict research, they have come with their own challenges. Most importantly, they often do not adequately consider the theoretical and conceptual implications of switching from conventional cross‐country to subnational analysis; this has led to dubious theoretical arguments and conclusions. Moreover, operationalization and measurement issues often limit these analyses’ explanatory power. The paper reviews several geospatial analyses of violent conflict, points out the limitations of the previous research and proposes some potential avenues for improvement.