Logo: to the web site of Uppsala University

This article presents new data on the direct and deliberate killings of civilians, called one-sided violence, in intrastate armed conflicts, 19892004. These data contribute to the present state of quantitative research on violence against civilians in three important respects: the data provide actual estimates of civilians killed, the data are collected annually and the data are provided for both governments and rebel groups. Using these data, general trends and patterns are presented, showing that the post-Cold War era is characterized by periods of fairly low-scale violence punctuated by occasional sharp increases in violence against civilians. Furthermore, rebels tend to be more violent on the whole, while governments commit relatively little violence except in those few years which see mass killings. The article then examines some factors that have been found to predict genocide and evaluates how they correlate with one-sided violence as conceptualized here. A U-shaped correlation between regime type and one-sided violence is identified: while autocratic governments undertake higher levels of one-sided violence than other regime types, rebels are more violent in democratic countries.

In many armed conflicts, rebel groups deliberately target civilians. This article examines whether such violence is related to the performance of the rebels on the battlefield. It is proposed that rebel groups who are losing battles target civilians in order to impose extra costs on the government. When rebels attack civilians, the government may incur both political and military costs. Violence against civilians is thus used as an alternative conflict strategy aimed at pressuring the government into concessions. The argument is evaluated by using monthly data for rebel groups involved in armed conflict from January 2002 to December 2004.