The status of amnesties under customary international law matters for two reasons. Firstly, customary international law is generally understood to create legally binding obligations for all states. This contrasts with international treaties, which only bind states that have ratified the treaties. Secondly, states have not (yet) created international treaties with duties to investigate, prosecute, and punish crimes against humanity and war crimes committed in non-international armed conflicts. Instead, these duties are argued to exist in customary international law. Assertions that amnesties are prohibited for crimes against humanity and war crimes in civil wars are similarly grounded in assertions that customary international law has evolved to oblige states to refrain from amnesties for these crimes.

This naturally raises the question of how we know if a rule of customary international law exists. The most authoritative answer comes from Article 38(1) of the Statute of the International Court of Justice. This Statute states that customary international law should be identified from two elements: state practice and opinio juris (a Latin phrase that means that the state believes that its practice was required by a legal obligation). The International Court of Justice has repeatedly emphasised this two-element approach in its case law. In doing so, it has found that for a rule of customary international law to exist, state practice must be ‘extensive and virtually uniform’ or ‘in general, be consistent with such rules…’ This approach suggests that if we want to identify a rule of customary international law then we need to observe what states are doing and saying in relation to the purported rule and we must determine if there is sufficient evidence to say that most states are willing to be bound by the purported rule.

However, to date, few international courts have sought to systematically observe state practice on this question. This includes the European Court of Human Rights, whose position on amnesties is of the greatest relevance for the ongoing domestic and Strasbourg litigation on the Northern Ireland Troubles (Legacy and Reconciliation) Act 2023.

Case-law

The Grand Chamber of the European Court of Human Rights in its 2014 Margus v Croatia case analysed the permissibility of amnesties under the European Convention on Human Rights in light of developments in international law. In doing so, it observed ‘[a] growing tendency in international law is to see such amnesties as unacceptable because they are incompatible with the unanimously recognised obligation of States to prosecute and punish grave breaches of fundamental human rights’ (para 139). However, the Court did not provide evidence of widespread state practice and opinio juris to support this interpretation; it relied instead on citing its own case law, the case law of other human rights bodies, and resolutions of international organisations. This evidential gap is significant as existing empirical data gives us reason to question whether there is indeed a ‘tendency’ in state practice to view amnesties for serious human rights violations as unacceptable. Indeed, looking at state practice in treaty negotiations, in adopting amnesties for serious human rights violations, and in responding to amnesties granted by other states, suggests that there is little state support for such a customary international law prohibition on amnesties.

Proposals to create treaty language to regulate or restrict when states can grant amnesties for serious violations were considered in the negotiations leading to the Convention on the Non- Applicability of Statutory Limitations to War Crimes and Crimes Against Humanity 1968; the Additional Protocol II to the Geneva Conventions 1977; the Rome Statute of the International Criminal Court 1998; the International Convention on Enforced Disappearances 2010; and the draft articles on a Convention on Crimes Against Humanity which are currently being developed by the UN General Assembly’s Sixth Committee. In each of these negotiations, proposed treaty provisions to bar amnesties could not gain consensus among states. As a result, no international treaty explicitly prohibits amnesties. On the contrary, Article 6(5) to Additional Protocol II encourages states to grant the broadest possible amnesty at the end of hostilities.

State practice

In addition, state practice suggests that that states continue to grant amnesties, even for serious violations. For example, the PA-X Peace Agreement Database contains information on 117 peace agreements reached between 1990 and 2024 with provisions committing the parties to grant amnesty for combatants. Only 29 of these agreements stated that amnesty could not be granted for genocide, war crimes or crimes against humanity. These 29 agreements related to only 16 countries. In addition, the Amnesties, Conflict and Peace Agreement database, which I created, contains information on 196 conflict-related amnesties relating to international crimes granted around the world between 1990 and 2023. 137 amnesties (70%) of these were granted for international crimes and serious human rights violations. In contrast, only 82 amnesties (42%) contained exclusions for these types of crimes. Some amnesty laws included some international crimes or human rights violations, while excluding others.

Finally, the Universal Periodic Review (UPR) provides us with a useful source of evidence of how states respond to amnesties granted by other states. The UPR is a cyclical peer review process led by the UN Human Rights Council, in which states make recommendations to other states on how to improve their human rights record. These recommendations are not legally binding, instead this process is intended to provide political pressure on states. From 2008 to 2021, states made 98,274 UPR recommendations to each other. However, only 25 of these recommendations called on states to refrain from amnesty or to review, amend or annul past amnesty. And these 25 recommendations related just to 14 countries. The very low numbers of recommendations calling on states to refrain from or annul amnesties for international crimes and serious human rights violations is indicative of how states generally greet amnesties enacted in other states with silence, or even support, rather than condemnation.

Overall, this data makes clear that state practice does not (yet) support a rule of customary international law prohibiting amnesties. There may be many reasons for this, but where states have granted amnesty or argued in international bodies that international law should not create an absolute prohibition on amnesties on international crimes, they have often emphasised the role that amnesty can play in ending conflict and promoting reconciliation. In addition, conditional amnesties can be used to facilitate offenders’ contributions to truth recovery or reconciliation. In the Margus case, the European Court of Human Rights appeared to acknowledge the potential positive outcomes of amnesties when it stated ‘Even if it were to be accepted that amnesties are possible where there are some particular circumstances, such as a reconciliation process and/or a form of compensation to the victims…’

Acknowledging the absence of customary

international law duty on states to refrain from amnesties does not mean that all states are required to grant amnesty. Nor does this mean that all amnesties would be legally permissible. Instead, it means that the legal permissibility of amnesties should be determined on a case-by-case basis through taking into account how the design of specific amnesties aligns to the enacting states’ treaty and domestic law obligations.