We examine how natural resource location, rent sharing and fighting capacities of different groups matter for ethnic conflict. A new type of bargaining failure due to multiple types of potential conflicts (and hence multiple threat points) is identified. The theory predicts conflict to be more likely when the geographical distribution of natural resources is uneven and when a minority group has better chances to win a secessionist rather than a centrist conflict. For sharing rents, resource proportionality is salient in avoiding secessions and strength proportionality in avoiding centrist civil wars. We present empirical evidence that is consistent with the model.