This paper examines the relationship between ethnic fractionalization, polarization, and conflict. While many studies have found that ethnic fractionalization negatively impacts economic growth, empirical evidence linking it to civil conflicts is weak. The authors argue that the channel through which ethnic fractionalization affects growth is not via increased potential ethnic conflict. They propose a new measure of ethnic heterogeneity, the Q index, which captures the concept of polarization better than traditional indices of fractionalization. The empirical results show that the Q index is a significant explanatory variable in the incidence of civil wars, robust to various alternative indicators of ethnic heterogeneity, data sources, and data types. The paper discusses the theoretical and empirical justifications for the Q index and compares it with the fractionalization index. It also explores the robustness of the findings to different definitions of civil wars, regional effects, and data sources. The results suggest that ethnic polarization, not fractionalization, is a stronger predictor of civil conflicts.This paper examines the relationship between ethnic fractionalization, polarization, and conflict. While many studies have found that ethnic fractionalization negatively impacts economic growth, empirical evidence linking it to civil conflicts is weak. The authors argue that the channel through which ethnic fractionalization affects growth is not via increased potential ethnic conflict. They propose a new measure of ethnic heterogeneity, the Q index, which captures the concept of polarization better than traditional indices of fractionalization. The empirical results show that the Q index is a significant explanatory variable in the incidence of civil wars, robust to various alternative indicators of ethnic heterogeneity, data sources, and data types. The paper discusses the theoretical and empirical justifications for the Q index and compares it with the fractionalization index. It also explores the robustness of the findings to different definitions of civil wars, regional effects, and data sources. The results suggest that ethnic polarization, not fractionalization, is a stronger predictor of civil conflicts.