Introduction
Civil wars show a remarkable variation in how long they last. A survey of all civil wars
begun in the last sixty years reveals that, while a quarter of the wars end within the first two
years, another 25% continue for longer than thirteen. Moreover, 10% of all civil wars last more
than two decades. What explains this wide variation in the duration of civil wars? Why are some
civil wars resolved quickly while others drag on for decades without resolution?
Civil war duration is an important topic for two main reasons. Theoretically,
understanding how and why civil wars are resolved through negotiated settlement inherently
involves an analysis of why it takes so long for settlements to be reached. If you want to
understand what causes combatants to fight and what gets them to cooperate, you must
understand the duration of these conflicts. The existing literature has largely ignored the duration
of civil war and looked simply at the outcome of these conflicts, which truncates a long and
important process.
Practically, the costs of war tend to be directly related to its duration. New data on battle
deaths in civil war reveals that internal conflicts that are two years or shorter generate, on
average, under 3000 deaths while civil wars that are longer than two years result in, on average,
over 44,000 deaths.
For each additional month of civil war, on average, an additional 189 people
are killed.
Understanding the factors that drive duration can help the international community
resolve conflicts early and save lives.
In recent years, the international community has recognized these costs and shown
greater interest in trying to resolve or at least put a lid on long-running civil wars such as those in
southern Sudan, Afghanistan and Liberia. Much of this behavior is driven by self-interest, as the
1
These estimates are from data on battle deaths collected by Lacina and Gleditsch 2004 for the Peace Research
Institute of Oslo/Uppsala Armed Conflict Database. The median duration in the database in 24 months, and reported
above are the mean number of battle deaths for conflicts that lasted up to twenty-four months or more than twenty-
four months.
2
This estimate is the result of a simple Ordinary Least Squares regression with Battle deaths as the dependent
variable and duration, in months, as the independent variable.
Convention