Many civil wars are struggles in which one or more ethnic factions claim that they seek to restore injustices imposed upon them by some other ethnic faction. This does not necessarily mean that ethnic grievances cause civil conflict. Indeed, much political science research argues that ethnic grievance are ubiquitous. Therefore, policy makers and researchers should focus on the factors that allow latent and peaceful conflicts to develop into manifest and violent ones rather than on rectifying perceived injustices.
Lars-Erik Cederman has a very interesting post on this issue at PoliSciZurich. The evidence for the ubiquity-of-grievances thesis is tentative as it relies on poor measures of grievances. Cederman and his research group have introduced a new set of geocoded measures of political access and inequality. Based on these measures, Cederman, Kristian Gleditsch, and Nils Weidmann find that disadvantaged groups are more likely to be involved in violent conflict. This is not my area of expertise but the entire post is well worth a read as it contains a good discussion of the issues, data, and policy implications.